Geochimica et Cosmochimica Acta, Vol. 68, No. 4, pp. 841–865, 2004 Copyright © 2004 Elsevier Ltd Pergamon Printed in the USA. All rights reserved 0016-7037/04 $30.00 ϩ .00 doi:10.1016/j.gca.2003.07.009 Volcanic arc of : a province with high-␦18O magma sources and large-scale 18O/16O depletion of the upper crust

1, 2 3 1 ILYA N. BINDEMAN, *VERA V. PONOMAREVA, JOHN C. BAILEY, and JOHN W. VALLEY 1Department of Geology and Geophysics, University of Wisconsin, Madison, WI, USA 2Institute of Volcanic Geology and Geochemistry, Petropavlovsk-Kamchatsky, 3Geologisk Institut, University of Copenhagen, Copenhagen, Denmark

(Received March 20, 2003; accepted in revised form July 16, 2003)

Abstract—We present the results of a regional study of oxygen and Sr-Nd-Pb isotopes of to Recent arc volcanism in the and the Kuriles, with emphasis on the largest - forming centers. The ␦18O values of phenocrysts, in combination with numerical crystallization modeling (MELTS) and experimental fractionation factors, are used to derive best estimates of primary values for ␦18O(magma). Magmatic ␦18O values span 3.5‰ and are correlated with whole-rock Sr-Nd-Pb isotopes and major elements. Our data show that Kamchatka is a region of isotopic diversity with high-␦18O basaltic magmas (sampling mantle to lower crustal high-␦18O sources), and low-␦18O silicic volcanism (sampling low-␦18O upper crust). Among one hundred and Late Pleistocene eruptive units from 23 volcanic centers, one half represents low-␦18O magmas (ϩ4 to 5‰). Most low-␦ 18O magmas are voluminous silicic related to large Ͼ10 km3 caldera-forming eruptions and subsequent intracaldera and domes: Holocene multi-caldera , and -Iliinsky , and Late Pleistocene Maly Semyachik, Akademy Nauk, and Uzon calderas. Low-␦18O magmas are not found among the less voluminous products of eruptions and these volcanoes do not show drastic changes in ␦18O during their evolution. Additionally, high-␦18O(magma) of ϩ6.0 to 7.5‰ are found among and basaltic andesites of , , , and volcanoes, and and of Opala and Khangar volcanoes (7.1–8.0‰). Phenocrysts in volcanic rocks from the adjacent Kurile Islands (ignimbrites and lavas) define normal-␦18O magmas. The widespread and volumetric abundance of low-␦18O magmas in the large landmass of Kamchatka is possibly related to a combination of near-surface volcanic processes, the effects of the last glaciation on high-latitude meteoric , and extensive geyser and hydrothermal systems that are matched only by Iceland. Sr and Pb isotopic compositions of normal and low-␦18O, predominantly silicic, volcanic rocks show negative correlation with ␦18O, similar to the trend in Iceland. This indicates that low-␦18O volcanic rocks are largely produced by remelting of older, more radiogenic, hydrothermally altered crust that suffered ␦18O-depletion during Ͼ2 My-long Pleistocene glaci- ation. The regionally-distributed high-␦18O values for basic volcanism (ca. ϩ 6toϩ7.5‰) in Kamchatka cannot be solely explained by high-␦18O slab fluid or melt (Ϯ sediment) addition in the mantle, or local of hydrated OIB-type crust of the Hawaii-Emperor chain. Overall, Nd-Pb isotope systematics are MORB-like. Voluminous basic volcanism (in the Central Kamchatka Depression in particular) requires regional, though perhaps patchy, remobilization of thick (30–45 km) Mesozoic- arc roots, possibly resulting from interaction with hot (ca. 1300°C), wedge-derived normal-␦18O, low-87Sr/86Sr basalts and from dehydration melting of lower crustal metabasalts, variably high in ␦18O and 87Sr/86Sr. Copyright © 2004 Elsevier Ltd

1. INTRODUCTION record are now attributed to sources in Kamchatka and the Aleutian island arc (e.g., Zielinski et al., 1996; Braitseva et Dynamic interplay between glaciations and volcanic activity is seen during the Pleistocene history of the northern circum- al., 1997b). The Kamchatka Peninsula overlies the northwest- Pacific region. Voluminous caldera-forming eruptions, glacial ern margin of the subducting Pacific plate and is one of the advances, and rapid sea level changes, have affected local and most volcanically active area in the Holocene (Simkin and global ; recent studies recognize the importance of the Siebert, 1994), and probably has the largest exposed hydrother- North Pacific in paleoclimate reconstruction (e.g., Bradley, mal and geyser system at high latitudes. 2000; Gualtieri et al., 2001; Beget, 2001; Brigham-Grette, This paper was prompted by the discovery that Kamchatka is 2001). Glaciation may have local and regional effects on vol- a land of isotopic diversity with high-␦18O basic magmas and canism; rapid pressure release during deglaciation may help low-␦18O upper crust. Low-␦18O magmas are here defined as trigger eruptions (e.g., Nakada and Yokose, 1992), but the Ͻ5.5‰ for basalts to Ͻ 5.8‰ for rhyolites; high-␦18O magmas broader effects are poorly understood. Many ash layers and are Ͼ6.0‰ for basalts, and Ͼ6.4‰ for rhyolites, see section

H2SO4-rich horizons in the Pleistocene-Holocene 4.1 below. Evidence is presented that Kamchatka is a high- latitude volcanic province with an abundance of low-␦18O rocks (based on number of low-␦18O units and their total * Author to whom correspondence should be addressed, at Division of Geological and Planetary Sciences, California Institute of Technology, volume) matched only by Iceland. We attribute this to the effect Pasadena, CA 91125, USA ([email protected]). of Pleistocene glaciation on meteoric waters, hydrothermally- 841 842 I. N. Bindeman et al. altered rocks, and magmas (cf. Donnely-Nolan, 1998; Gautason create a framework of Holocene volcanic activity, and Muehlenbachs, 1998; Bindeman et al., 2001). which is pertinent to volcanology, dating marine sediments No low-␦18O magmas have been found in low-latitude arcs (Prueher and Rea, 2001a), impact on climate (Braitseva et al., of the circum-Pacific, or other island arcs, but normal- and 1997b; Prueher and Rea, 2001b), and archeology (Braitseva et high-␦18O magmas are common (Taylor, 1968; Matsuhisa, al., 1987). Pleistocene and older and layers 1979; Macpherson et al., 1998; Vroon et al., 2001). At high- are unambiguously found in the vicinity of large calderas (Ͼ10 latitudes, the Fisher and Okmok calderas (Aleutians) produced km in diameter): Pauzhetka, Uzon, Maly Semyachik, Akademy low-␦18O magmas (Bindeman et al., 2001), and more low-␦18O Nauk, and others (e.g., Grib and Leonov, 1993a,b). Most te- magmas are likely to be found there: like Kamchatka, the phras at each eruptive center represent differentiated products, Aleutian arc is a volcanically active subpolar area that under- typically silicic andesite to , occurring in caldera set- went Pleistocene glaciation. tings, and are distinct mineralogically and chemically. Given This study uses the primary range of magmatic ␦18O values the predominantly NE direction of winds, tephra layers from in Kamchatkan magmas based on analyses of phenocrysts in major eruptions have been deposited in the Beringia Terrain combination with the radiogenic isotopes of Sr, Nd, and Pb (Fig. 1, inset), and numerous layers of abundant Kamchatkan determined on whole rocks to provide key constraints on the are identified in ODP marine sediments in the NW genesis of both high- and low-␦18O magmas discovered in Pacific (Bailey, 1993, 1996; Prueher and Rea, 2001a,b). Kamchatka. Such an approach is used to define the role of deep (slab and mantle) processes vs. shallow processes of crustal 2.2. Glaciations recycling, and was used in the past to describe more subdued ␦18O ranges in other island arcs and other settings (e.g., Thirl- Kamchatka experienced several voluminous glaciations dur- wall et al., 1996; Baker et al., 2000; Eiler et al., 2000a,b; Vroon ing Pleistocene (Braitseva et al., 1968; Grosswald, 1998). Ice et al., 2001). We here present evidence that lower and upper covered most volcanic edifices even in southern Kamchatka crustal processes dominate the petrogenesis of basic and silicic (Melekestsev et al., 1974; Savoskul, 1999). Starting from the magmas in Kamchatka, though a mantle signature can still be beginning of glaciation at 2.6 Ma, ice repeatedly reached sea- recognized in some trace elements and isotopes. shores and supplied debris and iceberg-rafted material to ma- rine sediments (Prueher and Rea, 2001b). At present, the snow ϳ 2. KAMCHATKAN VOLCANISM line is at 1000m elevation, and many volcanic edifices con- tain patchy glaciers. 2.1. Eruptive Centers, Volcanic Rocks, and 14C-Dated Tephra Layers 2.3. Previous Stable Isotope Studies

The Kamchatka is a part of the Kurile-Kam- Phenocryst-based oxygen isotope studies in Kamchatka have chatka volcanic arc related to the subduction of the Pacific plate been limited; most reported analyses are for whole-rocks that (Fig. 1). Eruptive centers differ in their tectonic position rela- are prone to postmagmatic alteration, and concentrated on the tive to the present volcanic front and can be grouped into three petrogenesis of basalts. Pokrovsky and Volynets (1999) and tectonic zones (Fig. 1): Eastern Volcanic Front (EVF), Central Pineau et al. (1999) demonstrated a moderate (ϳ1‰) increase Kamchatka Depression (CKD), and (SR). in whole-rock ␦18O values across the arc. Dorendorf et al. Magmatic volumes generally decrease westward, but Recent (2000) discovered that olivines and pyroxenes from Klyuchev- volcanism is most voluminous in the CKD with some of the skoy volcano have high-␦18O values (6–7.5‰) and attributed largest volcanoes of the world such as Klyuchevskoy. Volcanic this to the addition of high-␦18O slab fluids to the mantle rocks of the Kurile-Kamchatka island arc span compositions generation zones. In the study of differentiated products, Pok- from to rhyolite, from tholeiitic to calc-alkaline, and rovsky and Volynets (1999) noted that many ignimbrites in from low- to high-K and shoshonitic series (Fig. 2; Volynets, Kamchatka are characterized by 18O-depleted whole-rock val- 1994; Bailey, 1996). Olivine-phyric basalts are relatively rare, ues, which result from secondary hydrothermal alteration or and typical silicic rocks tend to lack quartz and zircon. The magmatic assimilation of hydrothermally-altered rocks. We majority of rocks are either basaltic andesites or andesites. This have compiled published whole-rock ␦18O values for Kam- is a common feature of island arcs, and regional oxygen isotope chatka and plotted them vs. 87Sr/86Sr for the same volcanic studies must rely primarily on the analyses of ubiquitous pla- units (Fig. 3A). The ␦18O values for basalts and basaltic gioclase, pyroxene, and amphibole phenocrysts. andesites, based on the petrographically freshest and least hy- Ͻ Unlike many previous studies, we emphasize the importance drated ( 1.1wt% H2O) whole-rock analyses, span the range of dated tephra and ignimbrites from products of major caldera- from 5.4 to 8.5‰, and 0.703 to 0.706 for 87Sr/86Sr (Ivanov, forming eruptions ranging from Pleistocene to Holocene and 1990; Pineau et al., 1999; Pokrovsky and Volynets, 1999). studied in detail during the last decade (Braitseva and Mele- Strontium and oxygen isotope studies of the exposed outcrops kestsev, 1990; Braitseva et al., 1995, 1996, 1997a,b; Leonov of Cretaceous to Miocene metamorphic basements (Fig. 3B) and Grib, 1998; Volynets et al., 1999; Ponomareva et al., demonstrate positive correlation, with higher 87Sr/86Sr and 2004). The 14C-dated Holocene tephras analyzed in this study ␦18O values for silicic metamorphic rocks of the Sredinny (Table A1, Electronic Annex, Elsevier Website, Science Di- Massif (SM, see Fig. 1). Geothermal waters in Kamchatka rect) are preserved regionally; they provide an important record (Vinogradov and Vakin, 1983; Taran et al., 1988; Chesho, of voluminous eruptions and serve as tephrochronological 1994) are low in ␦18O and ␦D, and seems to inherit these markers (Braitseva et al., 1995, 1997a). Identified regional relatively high 87Sr/86Sr values, even for geothermal systems Volcanic arc of Kamchatka 843

Fig. 1. Map of Kamchatka Peninsula showing Holocene volcanic centers, major Holocene and Late Pleistocene calderas, and regional geodynamic features after Bogdanov and Khain (2000) and Baranov (1991). The upper inset shows the direction of tephra deposition (Braitseva et al., 1997a) for most voluminous Holocene eruptions; tephra codes are from Table A1 (Electronic Annex). Lower inset shows global position of Kurile-Kamchatka arc and interconnected glacial covers from Mann and Peteet (1994) and Braitseva et al. (1968). Selected calderas and volcanoes of the Kurile island arc analyzed in this study (Table 1; and Table A1, Electronic Annex) are shown in the lower figure. G ϭ Golovnin caldera; M ϭ Mendeleyev volcano; BG ϭ Bogdan Khmelnitsky volcano; K ϭ Kudriavy volcano; Z ϭ Zavaritsky caldera; B ϭ Brouton volcano; A ϭ Alaid volcano. Open triangles and circles are studied volcanoes and calderas. Volcanism in Kamchatka occurs in three structural zones: the Eastern Volcanic Front (EVF), the Central Kamchatka Depression (CKD), and the Sredinny Range (SR). Following Late Miocene accretion of the Eastern Peninsula Terrains (Geist et al., 1994), the subduction trench rolled back 200 km to the east to its present position (Volynets, 1994) and the present position of the Kurile-Kamchatka volcanic front was established. Presently active volcanoes (Khangar, ) are located on the Cretaceous to Eocene siliceous metamorphic basement (Sredinny Massif, SM). To the east, mafic crust is exposed as high-grade metamorphic rocks of Ganal Massif (GM) of Cretaceous to Miocene age. Petrology, ages, and tectonic mechanisms of accretion are discussed in Geist et al. (1994); Tatsumi et al. (1995); Konstantinovskaya (2000); Bogdanov and Khain (2000) and Bindeman et al. (2002). 844 I. N. Bindeman et al.

Fig. 1. (Continued) Thus, recent volcanism is built upon a series of Mesozoic terranes with both felsic (SR) and mafic crusts (EVF, CKD,) with low 87Sr/86Sr (0.703 to 0.706, Vinogradov, 1995) that form 30 to 45 km thick crust (Bogdanov and Khain, 2000). In the NW Pacific and Kamchatka, nearly orthogonal subduction, dipping at ϳ45° angle of the Cretaceous Pacific plate for the past 43 Ma (e.g., Engebretson et al., 1985; Nokleberg et al., 1998) has enriched the mantle wedge with volatiles and, as is discussed in the text, has caused a regionally extensive high ␦18O source. Accreted oceanic arcs have added high-␦18Omafic and silicic rocks (Fig. 3B) to the lower crust/mantle. far from the surface outcrops of metamorphic basements; this 232Th vs. 238U/232Th diagram pointing that Ͼ150 k.y. has causes elevation of 87Sr/86Sr in 18O-depleted hydrothermally- elapsed since fluid release from the slab and eruption of mag- altered rocks (Fig. 3B). Values of ␦18O and ␦D of fumarolic mas (e.g., Turner et al., 1998); however voluminous basic rocks gases (Taran et al., 1988, 1997) indicate a significant proportion from the Central Kamchatka Depression show larger disequi- of meteoric . libria and may require “shorter” timescales (Dosseto and Bour- don, 2002); and 5) 10Be/9Be in Kamchatka are lower than in the 2.4. Radiogenic Isotope Studies Aleutians or the Kuriles (e.g., Tera et al., 1990; Ryan et al., 1995) suggesting longer magma residence times and/or deriva- Sr, Nd, Pb, Th, Os, Be, and B isotopic, as well as trace tion of magma from different mantle and crustal sources. elemental studies have been performed on Kamchatkan and Whereas mantle- and slab-derived processes are currently Kurile volcanoes, and sediments in the NW Pacific, and corre- interpreted to be the dominant control on radiogenic isotope lated to tectonic position and depth to the Benioff zone (Bailey et al., 1987; Zhuravlev et al., 1987; Tera et al., 1990; Bailey, systematics of Kamchatkan magmas, as in the Kuriles and 1993, 1996; Kersting and Arculus, 1994, 1995; Volynets, 1994; Aleutians, the Kamchatkan volcanism has some additional fea- Hochstaedter et al., 1996; Leeman, 1996; Kepezhinskas et al., tures. Voluminous volcanism, some with adakitic (high Sr/Y) 1997; Plank and Langmuir, 1998; Turner et al., 1998; Doren- affinities (Drummond et al., 1996) in the CKD is attributed to dorf et al., 2000; Ishikawa et al., 2001; Widom et al., 2003). melting of the Pacific slab due to mantle corner flow around the Most of radiogenic isotope variations are interpreted to indicate plate edge (Yogodzinski et al., 2001). Neogene subduction of that: 1) isotopic compositions of primary Kamchatkan lavas are the Meiji and Emperor Seamount Chain may cause an alkaline primitive and similar to Pacific MORB; 2) there is a small imprint and high magma production rates in Klyuchevskoy and amount of sediment input into magma generation zones; 3) other volcanoes of CKD (Kersting and Arculus, 1994, 1995; there are across-arc and along-arc variations in isotopic param- Bailey, 1996). These features are shown to be more regional by eters and in Ce/Y, Ba/Th, La/Ta, B/Nb suggesting differences recent mantle tomography (Gorbatov et al., 2001); a wide in fluid release as a function of slab depth (e.g., Ishikawa et al., ocean-ward mantle plume with anomalous high heat flow heats 2001); 4) most samples plot near the equiline on the 230Th/ and thins the oceanic lithosphere before it is subducted under

Fig. 2. Major element composition of samples analyzed in this study K2O vs. SiO2. See Table A1 (Electronic Annex) for analyses. Dashed lines divides low-, medium-, and high-K volcanic arc series (Gill, 1981). CKD and SR lavas tend to be more K-rich than EVF lavas. Klyuchevskoy data are from Dorendorf et al. (2000). Volcanic arc of Kamchatka 845

Fig. 3. Correlation of ␦18O(whole rock) vs. 87Sr/86Sr. (A) Quaternary Kamchatkan volcanic rocks. Published ␦18O(whole rock) analyses indicate that shallow processes of isotopic exchange and alteration dominate over primary (source) characteristics. Whole-rock data from published sources indicate wide ranges due to high- and low-temperature exchange with meteoric waters, which affect ␦18O(whole rock) much more than 87Sr/86Sr ratios. In contrast, assimilation of older metamorphic Mesozoic crust of Sredinny Range preferentially changes 87Sr/86Sr (highest 87Sr/86Sr values are from lower Pleistocene Belogolovsky volcano on SR, Pokrovsky and Volynets, 1999). (B) ␦18O(whole rock) vs. 87Sr/86Sr in other crustal materials: mafic (Ganal terrane) and felsic (Sredinny Range) metamorphic rocks from the basement, hydrothermally altered rocks, and geothermal waters of modern hydrothermal systems. Inset shows ␦D vs. ␦18O in hydrothermal (100–300°C) waters (Vinogradov and Vakin, 1983) relative to present-day meteoric water line (MWL). Isotope data are compiled from Taran et al. (1988), Ivanov (1990), Volynets (1994), Vinogradov (1995), Zolotarev et al. (1999), Pokrovsky and Volynets (1999), and Dril et al. (1999).

Kamchatka. Additionally, slab detachments at 10 and 2 Ma onstrates a largely crustal origin for most silicic and some basic (Levin et al., 2003) may have contributed to the diffuse mono- magmatism, as has long been advocated in other island arc genetic magmatism with intraplate geochemical affinity that systems (e.g., Davidson, 1989). It indicates the importance of occurs predominantly in the back arc (Volynets, 1994). All recycling of older arc volcanic material in Recent volcanism, these mantle-derived magmas rise through thick 30–45 km and indirect influence of last glaciation on the isotopic signa- continental crust that exists beneath Kamchatka (Bogdanov and tures of magmatism. Khain, 2000). These observations suggest that Kamchatka is a geochemi- 3. ANALYTICAL TECHNIQUES cally complex arc system where mantle-derived magmatism is We analyzed phenocrysts by laser fluorination to determine primary tectonically diverse and driven by a variety of processes. In magmatic ␦18O values at the University of Wisconsin. Phenocrysts are contrast to previous studies and interpretations, this paper dem- shown to be unaltered by petrographic inspection, and by analysis of at 846 I. N. Bindeman et al.

least two minerals in a rock, by duplicate analyses of individual (1–2 same homogeneous mantle reservoir at different P-T-XH2O mg) phenocrysts (e.g., plagioclase), and calculation of magmatic min- conditions, or interaction with other reservoirs. This has par- eral-mineral fractionations. Analyses of plagioclase, olivine, clinopy- roxene, amphibole, and quartz phenocrysts allow derivation of mag- ticular relevance for volcanic arcs where the magmatic rocks matic values using experimental and empirical mineral-melt and are characterized by a significant range of SiO2 contents, and a mineral-mineral isotopic fractionations (see below and Appendix). number of variable ␦18O reservoirs (slab, mantle wedge, crust, Several petrographically fresh whole-rock powders for which mineral or fluids) are thought to contribute to petrogenesis. separates were unavailable were analyzed; their magmatic ␦18O values are treated with caution. Analyzed mineral aliquots were typically 1–2 In this work, we undertook modeling (MELTS) of fractional ␮ mg, yielding 10–30 mol of CO2; yields were 90–102%, and there is crystallization of several typical arc basaltic compositions of no correlation between yield, sample size, and ␦18O. Minerals were Kamchatka (high-Mg, high-Al, high-alkali, Table A2, Elec- pretreated with 5 torr of BrF5 overnight to remove any surface or water tronic Annex) to derive the magnitude and trajectory of ␦18O contamination. The CO laser fluorination lab yielded precise ( Ϯ 2 increase with differentiation as a function of pressure and water 0.1‰, 1 st dev) analyses, used BrF5 reagent, and conversion to CO2 gas following fluorination. Four to nine UWG-2 garnet standards (ϩ5.8‰, content. We have attempted similar crystallization modeling corresponding to NBS28 ϭ 9.6‰, Valley et al., 1995) were analyzed with Kurile Island arc compositions in the past (Bindeman and Ϯ during each analytical session, and yielded 5.75 0.10‰ (1 st dev). Bailey, 1999); this yielded results in agreement with the ob- Eleven NBS-28 quartz standards yielded a value of ϩ9.46 Ϯ 0.08‰ (1 st dev). Whole rock powders (ϳ2 mg) were analyzed one at a time served chemical compositions of melts and minerals (plagio- using an airlock chamber. A correction in the range of ϩ0.3‰ to clase in particular). Ϫ0.05‰ (ϩ0.15 to 0‰ for most days) was applied to account for Knowledge of melting relations and 18O/16O isotopic frac- day-to-day variations based on values of 5 to 7 UWG-2 garnets mea- tionations guided the choice of initial compositions and exper- sured during each analytical session. Thus, the overall analytical un- certainty on single measurements is better than Ϯ 0.10‰ (1 st dev). imental parameters to derive the most ‘extreme’ (or end-mem- Sr, Nd, and Pb isotopes were analyzed in the University of Copen- ber-case) isotopic effects. In particular, low pressures and more hagen by TIMS (VG Sector 54-IT) using standard preparation tech- oxidizing conditions promote early crystallization of olivine niques of acid digestion (HBr, HNO , and HF) and cation-exchange 18 3 and magnetite, minerals that have the most negative ⌬ O(min- chromatography. Sr-isotopic ratios were measured in the dynamic ␦18 mode; Rb interferences were corrected for mass fractionation to 87Sr/ eral-melt), and promote the largest increase in O(melt). 86Sr ϭ 0.1194. Reproducibility was tested by measuring standard Higher water pressures enhance the crystallization fields of NBS987 which yielded 87Sr/86Sr ϭ 0.710244 Ϯ 13 (1 st dev). Nd olivine and pyroxene, and shrink the fields of spinel and pla- isotope ratios were corrected for Sm interference and for mass frac- gioclase, whereas high dry pressure enhances the role of py- tionation 146Nd/144Nd ϭ 0.7219. The laboratory JM Nd standard gave 144Nd/144Nd ϭ 0.511108 Ϯ 13 (1 st dev). Pb isotopic composition was roxenes on the liquidus (e.g., Presnall et al., 1979). ␦18 fractionation-corrected on average by 0.12% per a.m.u., relative to Pb The trajectory of O increase with SiO2 can be calculated standard NBS981; repeated analyses of this standard yielded 206Pb/ as a small-increment, forward-step model given the identities, 204 ϭ Pb 16.890–16.895 in the course of this work. Selected samples proportions, and composition of minerals crystallized, the were also analyzed for major and trace elements (Table A1, Electronic Annex) by XRF (Philips PW1400) at the Geological Institute of the (changing) composition of melt, and temperature. Whereas University of Copenhagen. there are numerous experiments on mineral-mineral fraction- ations (see Chacko et al., 2001; Eiler, 2001, for references), 4. DIFFERENTIATION MODELING there are only a few more difficult CO2-equilibration experi- 4.1. Normal-␦18O, Low-␦18O, and High-␦18O Magmas ments on isotope fractionations involving melts of various composition (Stolper and Epstein, 1991; Palin et al., 1996; In this work we define normal-␦18O magmas as having an Matthews et al., 1998; Appora et al., 2003). Nevertheless, these oxygen isotope ratio consistent with derivation by crystal dif- experiments demonstrate that melts can be treated as mixtures ferentiation from basalt of an arc-mantle source (ϩ5.8 Ϯ of normative mineral components, and thus mineral-melt equi- 0.2‰). Since the present work considers rocks of different libria can be calculated for a variety of melt compositions. ␦18 SiO2 content, it is important to estimate how much O Conveniently, MELTS modeling provides realistic proportions variation can be attributed to fractional crystallization. There and compositions of minerals and melts, CIPW norms, and have been a number of attempts in the past to estimate and enables calculation of isotopic effects at each temperature and 18 parameterize ␦ O change with compositional differentiation fractionation step. 18 monitors, such as SiO2, MgO and Fe/Mg used to measure Thus, the particular trajectory of ␦ O change can be mod- increasing differentiation (e.g., Taylor, 1968; Anderson et al., eled as: 1971; Matsuhisa et al., 1973; Matsuhisa, 1979; Muehlenbachs and Byerly, 1982; Kalamarides, 1984; Eiler, 2001). There is an ␦18 ϭ ␦18 ϩ ͸ ͑͑ ⌬ ͑ ͒͒ overall agreement that differentiation from basalt to rhyolite O (melt)j ϩ 1 O (melt)j Xi * i T (1) produces a positive subper mil change in ␦18O melt, since the 18 bulk cumulate has variable but lower ␦ O than the residual where Xi are proportions of differentiating minerals at each ⌬ melt. However, there are disagreements on the absolute range step j, and i(T) are temperature-dependent and melt compo- (from 0 to Ͼ1‰) and trajectory of ␦18O(melt) change with sition-dependent, mineral-melt fractionation factors. In addi- tion, the influence of plagioclase composition on melt was SiO2. The modeling of ␦18O shift due to fractional crystallization is recorded, as the change from anorthite to albite produces the ϳ ⌬ becoming increasingly important since the improved analytical strongest ( 1‰) effect on Plag-melt(T) (e.g., Matthews et al., precision of oxygen isotope analysis (better than Ϯ 0.05– 1983; Palin et al., 1996); effects of Mg-Fe substitution in mafic Ͻ ⌬ 0.1‰), can now resolve the cause for small (ca. 0.5‰) minerals on i(T) were ignored as being very small (e.g., variations, whether due to differentiation, partial melting of the Hoefs, 1997). Volcanic arc of Kamchatka 847

Fig. 4. A-D. Isotopic effects associated with fractional crystallization of basaltic magmas: results of numerical crystal- lization modeling (MELTS) using starting typical basaltic compositions (high-Mg, high-Al basalts given in Table A2, Electronic Annex, from Volynets, 1994) and isotopic fractionations discussed in text. (A) Trends for high-Al and high-Mg ␦18 basalts (this study) compared with the compilation of published O vs. SiO2 trends. (B, C) Isotopic effects as a function of different conditions of crystallization and starting compositions. (D) ⌬(melt-Plag) fractionation as a function of anorthite content of plagioclase; note that a crossover occurs fairly early in differentiation history at 55–60% SiO2 content of the melt. Kinks are due both to phase appearance and calculation step spacing (typically 10°C). 848 I. N. Bindeman et al.

The results of modeling are presented on Figure 4, and 1989), small ⌬18O indicate equilibrium at magmatic tempera- described in more detail in the Appendix. Conclusions can be tures suggesting that ␦18O values of phenocrysts reflect summarized as follows. quenched magmatic conditions. There is a spread of Ͼ3‰ in ␦18 18 (1) There are a variety of O(melt)jϩ1 trajectories with magmatic ␦ O values for each mineral. Twenty eight plagio- increasing degree of differentiation (f%), but they average-out clase-clinopyroxene pairs yielded a fractionation of 1.18 Ϯ and yield smoothly decelerating (concave downwards curves) 0.37‰ (1st dev). Nineteen plagioclase-amphibole fraction- ␦18 increases of O with increasing SiO2 of the residual melt ations produced comparable fractionation 1.23 Ϯ 0.30‰ (1 st ␦18 (Fig. 4a). Calculated trends, here called normal- O magma dev). The ⌬18O(Cpx-Amph) can therefore be estimated to be arrays, plot in the middle, but towards the lower end of cited 0.05‰, and this agrees well with estimates by Kohn and Valley natural estimates, and agree well with recent results for ocean (1998) of 0.1 Ϯ 0.2‰ at 800°C. Such ⌬18O(Cpx-Amph) frac- islands (Harris et al., 2000; Eiler, 2001). tionation was assumed in calculations of ␦18O(melt) below. (2) Fractional crystallization or melting at high mantle pres- Nine plagioclase-olivine pairs yielded 0.72 Ϯ 0.33‰, in good sures (p Ͼ 15 kbar) and temperatures yields pyroxene-domi- agreement with data of Eiler et al. (2000b) on lavas from other nated assemblages and produces negligible effects on residual island arcs. Xenocrystic and protocrystic olivines that result ␦18O(melt) (Fig. 4B). Differentiation at midcrustal pressures from preeruptive disintegration of olivine-anorthite cumulate (1–8 kb) with olivine and anorthitic plagioclase on the liquidus nodules, and magma mixing (e.g., Izbekov et al., 2002), occur (typical for island arcs, e.g., Bindeman and Bailey, 1999) yields sporadically; and are excluded from the calculation of ␦18O(m- a maximum 0.3‰ increase. Differentiation of the initially less ⌬ magnesian magmas, such as shoshonites and high-Al basaltic elt). Each (min-melt) value is a function of melt composition, ␦18 ␦18 mineral composition, and temperature. We used O for py- andesites, yields nearly linear O vs. SiO2 trends, and up to 0.45‰ increase from basalt to rhyolite (Fig. 4C). roxene, amphibole, olivine, quartz, and plagioclase to indepen- ␦18 (3) ⌬18O(Plag-melt) changes from a small (Ϫ0.1‰) negative dently calculate O(melt), after the following procedure. ␦18 value to a progressively more positive (0.3–0.65‰) value as Derivation of the O value of melt (and magma) in equi- the melt becomes increasingly richer in normative quartz and librium with phenocrysts is a necessary and precise procedure, albite (Fig. 4D). The crossover, inferred from rocks (e.g., since whole-rock analyses of volcanic rocks often prove to be Taylor, 1968), occurs at variable compositions of plagioclase unreliable for inferring primary magmatic values. Coexisting ␦18 —An85 to An55 for various studied compositions and parame- mineral pairs allow estimation of O values of magma at ters—and at SiO2 contents of 55 to 60 wt%. After the cross- temperatures and melt compositions using the above experi- 18 over, the bulk cumulate assemblage becomes lower in ␦ O mental fractionation factors. Increasing SiO2 content of the than the melt, and this feature persists even after the appearance melt is accompanied by a temperature decrease from ϳ1200°C ␦18 ϳ of quartz. This explains the slightly faster O vs. SiO2 in- to 800°C. The decrease in anorthite content of plagioclase is crease for initially more differentiated starting compositions. from ca. An80 (basalts) to An30 (rhyolites), and there is a Magnetite, as a minor phase, produces only a small (ca. simultaneous, but less significant decrease in Mg/Fe ratio of Ͻ0.05‰) effect on ␦18O(melt), even in oxidizing conditions. pyroxenes and amphiboles from 0.8 to 0.6–0.5. Crystallization (4) The results of this modeling are in good agreement with modeling yielded SiO2, T, and An content of plagioclase (e.g., ␦18 the magnitudes and values of O(melt) vs. SiO2 and plagio- Fig. 4), and CIPW norms of melt. A robust linear parameter- clase crossover relationships of concurrently published incre- ization of T, SiO2, and An content of plagioclase was assumed ment method calculation by Zhao and Zheng (2003). to match the observed variations and results (Table A2 and Fig. ␦18 Given the above definition of a “normal- O differentiation A1 in Electronic Annex). Next, a relevant subpermil fraction- Ϯ Ϯ array” (changing from 5.8 0.2‰ for parental basalts to 6.1 ation factor was added (or subtracted) to measured ␦18O values ϳ 0.3‰ for rhyolites at 90% differentiation, Fig. 4A), any of Fe-Mg minerals, plagioclase, and quartz to calculate ␦18O ␦18 magma with lower values is here called low- O, and any value of melts, in accordance with SiO contents of each magma above the differentiation array is called high-␦18O. 2 composition that indirectly include temperature dependence. High-␦18O values cannot be produced by closed-system evo- An internal check of MELTS calculations is achieved by lution from a mantle-derived arc basalt, and are generally comparing calculated ␦18O(melt) with measured ␦18O(WR) in derived from high-␦18O source rocks (e.g., supracrustals), or fresh rocks (Table 1). The difference: ␦18O(calc.) Ϫ result from addition of a high-␦18O component (e.g., slab or ␦18O(meas.) ϭ 0.02 Ϯ 0.34‰ (n ϭ 6) suggesting no intro- crustal fluids) to magma generation sources. Low-␦18O mag- ␦18 mas reflect assimilation or melting of low-␦18O rocks altered at duced bias in inferring O(melt) from phenocrysts. The re- ␦18 high-T by surface water. sulting O(melt), based on different minerals, plot within a range of Ϯ 0.2‰ and were averaged to obtain the preferred ␦18O(melt). Note that ␦18O(melt) is different from 4.2. Isotopic Fractionations and ␦18O of Melt in ␦18O(magma) since the latter is dependent on the % of crystals Equilibrium with Minerals and melt (i.e., magma ϭ meltϩphenocrysts). However, the proportion of crystals in Kamchatkan volcanic rocks is rela- Table 1 presents ␦18O values for phenocrysts from 111 tively small (Ͻ20 vol%), and ⌬(mineral-melt) fractionations eruptive units of 30 volcanic centers of Kamchatka and Kuriles. often show small, opposite signs. It is therefore safe to assume Sample descriptions and ages of units are given in Table A1 that for near-liquidus rocks ␦18O(melt) is equal to (Electronic Annex). The ␦18O values of coexisting minerals are ␦18O(magma), though possibly 0.1‰ higher than magma rep- plotted in Figure 5. On the basis of experiments (Chiba et al., resented by crystal-rich samples. Volcanic arc of Kamchatka 849

Table 1. Oxygen isotope composition of phenocrysts in Kamchatkan volcanic rocks. See Table A1 (internet repository) for description of eruptive units and whole-rock analyses. SiO2 and K2O given here are LOI-free.

␦18 87 86 SiO2 K2O O, permil Sr/ Sr

Sample wt% wt% Qz PI Amph Cpx Opx Mt OI Glass magma (calc)

Shiveluch 97044/1 62.44 1.53 7.30 5.68 7.14 97051/2 59.11 1.32 7.30 5.58 6.99 SH-3 57.04 0.99 7.08 5.53 6.80 96025/4 60.03 1.37 5.50 6.75 97049/2 57.11 1.10 5.83 6.99 1189/1 50.16 1.68 6.95 6.95 0.703673 OOK37 53.83 0.91 6.83 6.56 97058/2 53.40 0.96 7.26 5.71 6.75 5734 51.01 1.67 6.36 5.83 6.76 0.703703 5764 53.87 1.28 5.61, 5.68 6.08 0.70347 B544/N 47.23 0.72 7.49, 6.92 0.70341 B541/2v 54.71 1.19 6.56 5.87 5.27 6.05 0.70335 B547 60.95 1.30 7.15 5.72 6.64 0.70335 B548b 62.58 1.47 7.79 6.05 5.31 7.49 0.70341 Bezymianny B580 58.91 1.49 7.97 7.01 7.72, 8.29 7.88 0.70361 B574/A 56.00 1.09 7.27 0.70358 B-584 64.62 1.42 7.51 0.70359 B600 55.81 1.11 7.56 6.81 6.61 0.70354 TOL-2 50.85 0.88 5.63 5.52 5.69 6.22 0.70344 Teklentunup 7305 57.97 4.56 4.85 3.74 4.70 0.70366 1019/1 50.16 1.85 5.52 5.83 0.70366 Kizimen 80013/4 63.62 1.89 7.21 5.63 7.11 0.703307 K35/1 63.72 1.49 7.02 6.19 7.30 Khangar 98032/4 69.26 2.56 6.95 5.54 7.14 0.70343 98032/2 68.41 2.36 6.61 5.75 7.01 0.70338 Uzon 87L-103 66.10 1.57 4.73 3.66, 3.74 4.84 0.703297 99L-101 68.06 2.00 4.67 3.10 4.59 0.703346 D493-74 72.24 3.52 4.98 4.57, 4.29 5.61 0.703494 D388i-74 71.62 2.17 4.88 5.14 0.703369 D368-74 71.74 1.98 4.44 3.13 4.72 0.703357 W-11 65.99 1.30 4.75 3.36 4.70 0.7035 Maly Semyachik 5230/1 66.78 2.00 4.49 3.20 4.51 4.55 0.703193 Bol Semyachik 90L-8 64.07 1.99 6.34 5.21 6.36 0.703443 84L-103 70.53 1.94 7.47 6.57 6.70 0.703444 Karymsky 808 KAR 4 67.26 2.45 5.02 3.88 5.13 0.703198 KRM 5ab-4 72.59 2.93 4.86 5.15 0.703181 99IPE-4 69.58 2.37 5.25 3.56 5.16 99IPE-8 62.30 1.56 5.36 4.32 5.37 99IPE-22a 52.38 0.59 4.84 4.27 4.48 5.36 4.85 99IPE-12 60.41 1.30 5.44 4.38 5.39 99IPE-6 69.54 2.45 4.92 4.01 5.22 J4257 75.93 3.37 3.42 3.42 Avachinsky 99201/1 60.66 0.92 7.06 5.95 7.22 0.703374 99163/9 53.02 0.70 7.10 6.00 7.04 0.703459 29135 57.22 0.70 5.61 6.78 29215 50.95 0.70 5.89 6.10 6.54 TOL-3 51.85 0.37 6.49 0.70335 Koryaksky 26061 50.12 0.91 5.85 6.16 26018 58.96 1.63 6.12 26129 55.20 1.50 5.82 (Continued) 850 I. N. Bindeman et al.

Table 1. (Continued)

␦18 87 86 SiO2 K2O O, permil Sr/ Sr

Sample wt% wt% Qz PI Amph Cpx Opx Mt OI Glass magma (calc)

Chasha Crater 98KAM2.3 72.63 3.46 7.50, 7.70 7.89 Opala Barany amphiteatre Crater 98 KAM 2.4 73.68 3.65 7.06 7.38 0.7033 Summit Crater 98-33/2 65.54 2.87 6.84 5.96 7.15 0.703179 98-10 73.64 3.55 7.34, 7.44, 7.46 7.70 Gorely 3667 64.77 2.64 4.44 5.22 5.44 0.70324 Mutnovsky B618/3 50.02 0.37 5.93 0.70338 B615/B 51.52 0.53 5.82 0.70315 B610A 59.49 1.69 5.7, 5.57 0.70338 B615A 51.52 0.53 6.43, 6.07 0.70329 Nachiki TOL-4 75.59 4.49 7.91 8.55 8.51 0.70365 Ksudach 8882/2 68.45 1.48 5.56 4.66, 4.56 5.80 0.703276 C999 54.18 0.62 5.15 4.21 4.97 0.703311 8880/5 71.78 1.57 4.68, 4.69 3.46 4.89 0.703389 C953 65.35 1.20 4.75 3.74 4.53 4.87 0.70337 86039/14 62.57 1.15 5.02 3.71 5.01 C918a 62.44 1.06 5.64 4.64 5.62 0.703312 8889/2 69.55 1.51 4.92 3.96 5.19 8889/3 62.79 1.14 5.12 5.11 Ks-alliw 50.00 5.20 4.63 5.33 0.70331 C977a 63.22 1.15 5.01 4.09 5.11 C977 64.75 1.16 4.94 4.99 0.703318 KA-38a/7 49.35 0.22 5.84 0.70331 KA110-7 58.68 0.77 5.41 0.70322 KA25A-7 67.15 1.20 5.09 0.70325 KA39/7 48.21 0.19 5.62 4.43 5.06 0.70339 KA1/7 56.79 0.64 5.74 0.70326 Pauzhetka 83L20 71.14 2.96 7.15 6.04 6.52 0.703255 Kurile Lake KAM-03a 73.82 1.90 5.73, 5.51 4.38 2.05, 2.06 5.90 0.703312 KAM-29AB 54.63 0.49 5.23 4.16 3.07 5.11 0.703251 KL-1 68.90 1.66 5.35 4.21 5.73 0.703347 KL-2 56.52 1.34 5.59 4.60 5.40 0.703281 97KAM-02 69.52 1.76 4.37 4.35 0.00 97KAM-11 72.81 1.86 5.41 4.21 0.00 97KAM-29B1 66.64 1.48 5.55 4.24 4.32 0.00 97KAM-29AL 71.38 1.91 5.43 4.12 0.00 86680 64.65 1.31 5.68 4.67 5.78 Iiyinsky Volcano 650 64.40 1.56 4.52 3.83 4.77 621 59.88 1.26 4.96 3.85 4.87 615 52.37 0.57 5.30 4.98 96K8-14 51.92 0.46 4.92 3.36 4.36 96K8-24 59.47 1.13 4.95 4.84 8645/1 63.41 1.47 5.00 5.01 Zheltovsky Volcano CI-114 50.00 5.72 5.24 5.43 Dikii Greben Volcano B319-1 65.11 1.49 7.42 5.85 6.15 0.70323 B318-1 67.45 1.69 7.29 5.85 6.15 0.70322 B333-1 52.71 0.87 6.09 5.78 0.70322 B319-6 64.79 1.38 5.63 5.68 0.70322 B325-1 54.77 0.82 7.31 5.96 0.70318 B325 65.30 1.45 7.35 6.32 0.70322 B334 70.63 2.20 7.27 6.40 0.70322 (Continued) Volcanic arc of Kamchatka 851

Table 1. (Continued)

␦18 87 86 SiO2 K2O O, permil Sr/ Sr

Sample wt% wt% Qz PI Amph Cpx Opx Mt OI Glass magma (calc)

Kurile Island Arc Golovnin Caldera (Kunashir) 416-4 64.57 0.42 6.82 4.54 5.70 G-117 64.57 0.42 4.93 6.09 116b 49.97 0.13 4.85 5.16 116-1a 49.97 0.13 5.10 5.02 5.57 Zavaritsky Volcano () ZAV-1 67.01 0.65 5.86 4.88 5.84 Mendeleev Volcano (Kunashir) 7587 56.32 0.20 6.76 5.75 7585 75.95 0.99 6.99 6.07 6.46 Kudriavy Volcano () B605 57.93 0.57 5.39 5.94 Bogdan Khmeinitsky (Iturup) B535 56.03 1.42 6.03 5.08 5.70 Alaid Volcano () 92-223 48.50 1.72 6.34 5.17 5.66 Brouton Volcano (Brouton) B15-307 58.22 2.10 6.06 5.02 5.29 5.92 0.70306

**Calculated using ␦18O values of phenocrysts and inferred mineral-magma fractionation (see text).

5. RESULTS 5.1.1. High-␦18O Basic Rocks of Kamchatka 5.1. ␦18O Values of Volcanic Rocks Basic rocks of Kamchatka are either normal-␦18O, or high- Table 1 presents best estimates of ␦18O(magma) in Kam- ␦18O (Fig. 6, Table 1) whereas low-␦18O basalts are absent chatkan volcanic rocks and these are used in combination with (Fig. 3). Dorendorf et al. (2000) reported high-␦18O values for other isotopic systems below. olivines and pyroxenes of Klyuchevskoy volcano, the largest

Fig. 5. ␦18O(Plag) vs. ␦18O(Cpx, Amph) showing the range of ␦18O values in Kamchatkan magmas. Note that isotopic fractionations are consistent with equilibration at magmatic temperatures. Experimental ⌬(Plag-Cpx) fractionations are from ⌬ Chiba et al. (1989) for An50 at 1200°C, and An30 at 700°C, (Plag-Hb) are 0.1‰ larger. Specified volcanoes are discussed in the text. Fields for high- and low-␦18O magmas are shown for comparison. Notice that low-␦18O samples are predominantly clinopyroxene-bearing, while high-␦18O samples are amphibole-bearing. 852 I. N. Bindeman et al.

␦18 ␦18 ␦18 Fig. 6. O(magma) vs. SiO2 content for ; the O(magma) is calculated from the O values of phenocrysts, see text. The thick line is a calculated normal-␦18O array, see Figure 4A. The shaded field shows normal-␦18O (mantle magma fractionation trend), and low-␦18O magmas (area with negative slope, Unimak and Umnak islands) for the Aleutian and Alaskan magmas (Bindeman et al., 2001); hatched line is a trend for Japanese volcanoes (Matsuhisa, 1979). Tonga-Kermadec data are within thin line area (I. N. Bindeman et al., unpublished data). Note that more than half of Kamchatkan volcanoes plot in the field of low-␦18O magmas, and other half in the field of high-␦18O magmas. Lower latitude circum-Pacific arcs are normal-␦18O.

volcano of Eurasia. The present work greatly expands the values in Kamchatka (ϳ4.5‰) are found in voluminous Uzon number of mantle-derived basalts of Kamchatka with a high- rhyodacitic ignimbrites and intracaldera volcanics (see Table ␦18O signature. High-␦18O basalts and basaltic andesites are 2). This value is 1.5–2‰ lower than the estimated value of ca. characteristic of Bezymianny, Shiveluch, Avachinsky, and Ko- ryaksky volcanoes. Additionally, we have analyzed a suite of Table 2. Robust linear approximations of ⌬(min.-melt), SiO (wr), 25 magnesian olivines (Fo90) from the most primitive basalts 2 from 14 volcanoes and monogenetic cones throughout Kam- T°C, and An% relations based on 75 compositions in six crystallization experiments (see Appendix for plotted data). chatka supplied to us by Dr. M. Portnyagin. The olivine ␦18O values range from 5.3 to 6.8‰, and individual results will be ⌬ (min.-melt) ϭ a[SiO2 wt%] ϩ b ␦18 reported elsewhere. High- O magmatic values in intermedi- ab ate and silicic rocks characterize differentiated products of Khangar, Opala, Chasha, and Kizimen volcanoes. Therefore, Clinopyroxene 0.061 Ϫ2.72 high-␦18O values are not constrained to the Central Kamchatka Olivine 0.088 Ϫ3.57 Ϫ Depression but also characterize volcanoes in the Eastern Vol- Plagioclase 0.027 1.45 canic Front (e.g., Avachinsky, Koryaksky, and Opala), and on the Sredinny Range (Khangar). ⌬ (min.-melt) ϭ c[T°C] ϩ d cd

18 5.1.2. Low-␦ O Silicic Magmas of Kamchatka Clinopyroxene Ϫ0.0033 4.16 Olivine Ϫ0.0048 6.42 In contrast to basalts, the studied intermediate to silicic Plagioclase Ϫ0.0016 1.75 rocks, as well as mafic cumulates and mafic rocks associated with some calderas, from 100 units of 23 volcanic centers have ⌬ (min.-melt) ϭ e[An%] ϩ f ␦18 ␦18 unusually low- O and plot below the normal- O fraction- ef ation trend (Fig. 6). Such a great abundance (by number of units and by volume) of low-␦18O magmas makes Kamchatka Clinopyroxene Ϫ0.024 2.37 Ϫ one of the largest Quaternary low-␦18O volcanic areas on Earth. Olivine 0.034 3.75 Plagioclase Ϫ0.010 0.78 Only Iceland has similar, large areas of low-␦18O volcanics (Muehlenbachs et al., 1974; Condomines et al., 1983; Taylor T(°C) ϭϪ17.2[SiO2 wt%] ϩ 2014; An% ϭ 0.128[T°C] Ϫ64.99; and Sheppard, 1986; Hemond et al., 1988). The lowest ␦18O An% ϭϪ2.435[SiO2 wt%] ϩ 206.25. Volcanic arc of Kamchatka 853

6.1 Ϯ 0.3‰ for normal-␦18O and rhyolites, but 3‰ 5.1.4. Major Holocene Calderas and Volcanoes of the Kurile lower than basic magmas from typical high-␦18O Kamchatkan Island Arc sources. We find that there is no regional or geodynamic control on We analyzed ␦18O in minerals from caldera-forming erup- the appearance of low-␦18O silicic magmas. Neighboring vol- tions of several volcanoes and the largest (Ͼ10 km) Holocene canoes often do not show any 18O-depletion (see Fig. 1, Table calderas (Zavaritsky and Golovnin) of the Kurile island arc ␦18 1). There is no Kamchatka-wide correlation of O with K2O (Table 1; Table A1, Electronic Annex). Silicic and basic rocks or SiO2 or any other chemical or isotopic parameters (see of historic and radiocarbon-dated eruptions—Mendeleyev, below). However, the specific values of these parameters, in Bogdan Khmelnitsky, Kudriavy, Brouton, and Alaid stratovol- these low-␦18O centers, provide a valuable discrimination of canoes—were also studied. In sharp contrast to Kamchatka, all different tephra layers and the products of major caldera- analyzed samples of the Kurile island arc, from caldera or forming eruptions. Most low-␦18O magmas tend to be pyrox- stratovolcano, represent normal-␦18O magmas. Radiogenic iso- ene-bearing and amphibole-free (Table 1, Fig. 5), suggesting tope systematics of the Kurile island arc (Bailey et al., 1987; lower water content in parental magmas, and/or remelting at Zhuravlev et al., 1987; Bindeman and Bailey, 1999) are con- shallower depths. High-␦18O magmas are, on the contrary, sistent with derivation of a compositionally-diverse basaltic predominantly amphibole bearing. This important observation suites (tholeiitic, calc-alkaline, and shoshonitic) from a mantle goes against the intuitive understanding that low-␦18O magmas source that is 1) changing from ultra-depleted to depleted; 2) should be richer in water, and thus lower in ␦18O. Skaergaard has negligible contribution from sediments (Tera et al., 1990; low-␦18O gabbros are not noticeably richer in hydrous minerals Ryan et al., 1995), and 3) has small contribution from slab (Taylor and Sheppard, 1986). fluids.

␦18 5.1.3. Calderas vs. Stratovolcanoes, and Low- O Magmas 5.2. ␦18O vs. Other Isotope Systems

The great majority of voluminous low-␦18O magmas appear Table 3 contains analyses of Sr, Nd, and Pb isotopes per- in calderas, and especially multicaldera volcanoes (Table A1, formed by us on the suite of studied samples; additional 87Sr/ Electronic Annex, and Table 1) varying in age from Early 86Sr ratios for the same units or the same samples are taken Pleistocene to historic (Sheymovich, 1979; Erlich, 1986; Fe- from the published sources (see Fig. 3B). The ␦18O(magma)- dotov and Masurenkov, 1991), though not all calderas contain radiogenic isotope correlations are plotted on Figures 8 and 9, ␦18 low- O magmas. Thus, it seems that the necessary condition and lead to following observations. ␦18 for normal- to high- O mantle-derived magmas to acquire a 1) Regardless of SiO content, Pb and Nd isotopic ratios in ␦18 2 low- O signature in the Kamchatkan crust is a shallow all studied Kamchatka volcanics, including the most differen- . The latter condition is characteristic of tiated products of caldera-forming eruptions (Tables 1 and 3), calderas; under stratovolcanoes, magma chambers are likely to plot within the range of normal Pacific MORB, as do Kurile be deeper, less evolved, and shorter-lived. Since large shallow island arc volcanics (Bindeman and Bailey, 1999). 143Nd/ magma chambers often contain silicic differentiates, the low- 144/Nd shows no trends with ␦18O (Fig. 9C) and there is a ␦18 O values characterize primarily silicic magmas. Addition- scattered relation between 143Nd/144/Nd, 206Pb/204Pb, 208Pb/ ally, a caldera is an important enclosed hydrogeological locus 204 207 204 Pb, Pb/ Pb and SiO2 (not plotted). In EVF, at Ksudach for the development of a geothermal system driven by a high and Kurile Lake, the more differentiated products are only heat flux. It seems that the existence of an older overlapping or slightly more radiogenic than mafic rocks (Bindeman and enclosing caldera (e.g., Ksudach volcano, Volynets et al., 1999) Bailey, 1994). A variability of 0.0002–0.0003 Sr units exists for which hosted an earlier hydrothermal system before the most the Khangar volcanic center (Dril et al., 1999), and is explained recent caldera event, is an important factor in the appearance of by assimilation of radiogenic metamorphic basement. 18 low-␦ O magmas. 2) 87Sr/86Sr ratios in mafic Kamchatkan volcanics are higher For two multicaldera volcanic centers—Pauzhetka-Kurile than those in Pacific MORB, and show an overall positive 14 Lake-Ilyinsky, and Ksudach—considerable C age data are correlation with ␦18O (Fig. 8B). The trend is particularly ap- available (Table A1, Electronic Annex). The ␦18O values for parent for normal- to high-␦18O rocks of CKD: Klyuchevskoy, phenocrysts and magmas when plotted against age exhibit Bezymianny, Zarechny, Kharchinsky, and Shiveluch volca- variations in ␦18O between different eruptive units of the same noes. Remarkably, there is a negative correlation of 87Sr/86Sr volcanic system (Fig. 7), suggesting a shallow origin of the with ␦18O for low- to normal-␦18O rocks, that tend to be silicic magmas and their low-␦18O values. Mafic cumulates, and the in composition (Figs. 8A,C). This trend is apparent for Kurile products of zoned silicic-mafic ignimbrite eruptions, have com- Lake, Ksudach, and Uzon, all in the EVF with a regionally parable levels of ␦18O depletion in the first-erupted silicic similar 87Sr/86Sr ratio in parental basalts. A similarly negative portions and in subsequent more mafic portions (Table 1), trend is observed in rift-derived tholeiites from Iceland; this similar to what is observed in the zoned - suggests that the ambient Icelandic crusts and Kamchatka’s are Fisher and Okmok eruptions in the Aleutians (Bindeman et al., older, more radiogenic, and are more depleted with respect to 2001). This implies that the hot basic magmas that initiated ␦18O (down to Ϫ10‰, Hattori and Muehlenbachs, 1982; He- melting or mixing had enough time to exchange oxygen in the mond et al., 1988, although bulk is probably 0 Ϯ 3‰). Lead upper crustal magma chamber. isotopic signatures for the same Kamchatkan eruptive units 854 I. N. Bindeman et al.

Fig. 7. Evolution patterns for two caldera complexes: Kurile Lake (A) and Ksudach (B). Data are from Table 1; see Table A1 (Electronic Annex) for description and 14C ages of eruptive units. Dashed lines delineate the field of normal-␦18O silicic

magmas. Olivines (Fo75) in Ksudach are xenocrysts derived from preeruptive disintegration of olivine-anorthite cumulates. See Erlich (1986), Macias and Sheridan (1995), Volynets et al. (1999), Rinkleff (1999), and Ponomareva et al. (2004) for geology and geochemistry of these caldera complexes.

(Figs. 9A,B) show a more subdued negative correlation, and 9D) Kamchatka volcanics point toward Pacific sediments; suggest that most assimilants are also slightly more radiogenic 206Pb/204Pb vs. 208Pb/204Pb and 207Pb/204Pb (Figs. 9E,F) plot with respect to lead isotopes. slightly above the Pacific MORB field, towards the Indian (3) Dredged sediments in front of the Kamchatka arc (silicic MORB. MORB-like Pb isotopic signatures (e.g., Kersting and ooze, altered basalts, and mud) have expectedly high ␦18O(9to Arculus, 1995), and other geochemical monitors of Kamchat- 23‰), and 87Sr/86Sr values (Table 4), and also show positive kan volcanics, are consistent with negligible sediment-control correlation. On bulk average, however, both O and Sr isotope (Ͻ1–2% sediments) on Sr and O isotopic systematics in vol- values are only moderately high in sediments. Other isotope canics. At face value, the more primitive compositions of Pb analyses of sediments in NW Pacific (see Bailey, 1993, 1996; and Nd in CKD suggest smaller involvement of slab sediments Kersting and Arculus, 1995; Plank and Langmuir, 1998) are in the petrogenesis of CKD magmas (e.g., Churikova et al., similar to our near-Kamchatka values. 2001); elevated values of 87Sr/86Sr in voluminous volcanism in (4) Radiogenic isotopes when plotted against each other CKD can be interpreted to represent fluid addition with small highlight regional differences between volcanics in the CKD concentration of Pb and Nd (e.g., Dorendorf et al., 2000). We and EVF. CKD volcanics are least radiogenic with respect to note here that the isotopic analyses of this study, as well as Pb and Nd isotopes, but span an equal or even somewhat larger those reported for predominantly mafic volcanics by Churikova range in Sr and O isotopes. On 87Sr/86Sr vs. 143Nd/144/Nd (Fig. et al. (2001) and Kepezhinskas et al. (1997), overlap and Volcanic arc of Kamchatka 855

Table 3. Whole-rock radiogenic isotope analyses of selected Kamchatkan volcanic units. See Tables 1, and A1 for description.

87 86 206 204 207 204 208 204 143 144 Volcano SiO2 (WR) Sr/ Sr Pb/ Pb Pb/ Pb Pb/ Pb Nd/ Nd

Shiveluch B544/N 47.23 0.70341 18.258 15.491 37.901 B541/2v 54.71 0.70335 18.331 15.509 37.945 0.513140 B547 60.95 0.70335 18.362 15.493 37.932 B548b 62.58 0.70341 18.304 15.490 37.885 0.513097 Bezymianny B580 58.91 0.70361 18.057 15.502 37.753 B574/A 56.00 0.70358 18.251 15.488 37.862 0.513090 B-584 64.62 0.70359 18.267 15.491 37.891 0.513100 B600 55.81 0.70354 18.261 15.496 37.935 Tolbachik TOL-2 50.84 0.70344 18.356 15.564 38.035 0.513075 Uzon 87L-103 61.10 0.703297 18.376 15.515 38.095 0.513071 99L-101 67.53 0.703346 18.376 15.515 38.099 0.513066 D493-74 72.10 0.703494 18.332 15.501 38.033 0.513067 D388i-74 71.40 0.703369 18.367 15.520 38.110 0.513070 D368-74 70.96 0.703357 18.361 15.507 38.075 0.513072 W-11 65.99 0.703500 Maly Semyachik 5230/1 66.78 0.703193 18.389 15.498 38.051 0.513111 Bolshoi Semiachik 90L-8 63.82 0.703443 18.365 15.513 38.091 0.513035 84L-103 70.64 0.703444 18.379 15.501 38.067 0.513065 Avachinsky TOL-3 51.85 0.703350 18.427 15.526 38.155 0.513070 Mutnovsky B618/3 50.02 0.70338 18.358 15.520 38.141 0.513113 B615/B 51.52 0.70315 18.333 15.498 38.073 B610A 59.49 0.70338 18.350 15.500 38.110 0.513060 B615A 51.52 0.70329 18.340 15.507 38.099 Nachiki TOL-4 75.59 0.70365 0.513028 Ksudach C977 62.99 0.703318 18.319 15.496 38.093 KA-38a/7 49.35 0.70331 18.350 15.521 38.163 0.513070 KA110-7 58.56 0.70322 18.043 15.585 38.098 KA25A-7 67.15 0.70325 18.349 15.500 38.117 0.513063 KA39/7 48.21 0.70339 18.568 15.528 38.091 KA1/7 56.79 0.70326 18.095 15.572 37.945 Pauzhetka 83L20 70.34 0.703255 18.358 15.504 38.141 0.513056 Kurile Lake KL-1 68.94 0.703347 KL-2 56.52 0.703281

2␴ uncertainties: 87Sr/86Sr: Ϯ 0.00001; 206Pb/204Pb and 207Pb/204Pb: Ϯ 0.01; 208Pb/204Pb; Ϯ0.02; 144Nd/143Nd: Ϯ0.000005.

demonstrate that regardless of SiO2 content they also lie in the (e.g., Bindeman and Bailey, 1999, and references therein). field of lower and upper crustal sources of Mesozoic to Mio- When plotted against crustal thickness (Fig. 10 A,B) 87Sr/ cene age. 86Sr and ␦18O exhibit significant scatter; mafic volcanics in the EVF show a scattered positive correlation with crustal 5.3. Correlation with Crustal Thickness and Slab Depth thickness, whereas silicic volcanics from the same volcanoes and crustal thickness lose this trend. Similar poor trends are Isotopic compositions of oxygen and strontium when plot- observed in isotopic variations of Nd and Pb (not plotted; see ted against the slab depth (Figs. 10C,D) demonstrate an Table 3 and Fig. 9). overall positive correlation for Kamchatka (with the excep- Radiogenic isotope ratios for mafic volcanics (Figs. 8, 9) tion for Shiveluch which is built on a thinned crust). This suggest that volcanic rocks of the CKD are significantly dif- departs from the negative correlation of 87Sr/86Sr vs. depth, ferent from any other rocks in the Kurile-Kamckatka arc, and and no dependence for ␦18O vs. depth, in the Kurile island form a separate (decoupled) field on any diagram. We also note arc, a southern extension of the same subduction zone that is that many of the across-arc trends reported in Dorendorf et al. built upon far thinner crust (20–30 km); the petrogenesis (2000) and Churikova et al. (2001) are possibly related to this there is interpreted to be dominated by mantle processes regional difference and may reflect the strikingly different 856 I. N. Bindeman et al.

␦18 87 86 ␦18 Fig. 8. O(magma) vs. Sr/ Sr and SiO2 correlations for Kamchatkan volcanic rocks; O(magma) values are calculated from phenocrysts (see text and Table 1). (A) All data; notice V-shaped overall correlation: there is a negative correlation of ␦18O(magma) vs. 87Sr/86Sr for low-␦18O magmas from volcanoes of the EVF (Kurile Lake-Ilyinsky, Ksudach, and Uzon), and an overall positive correlation of ␦18O(magma) vs. 87Sr/86Sr Kamchatka-wide from EVF to CKD. Mass balance curves for bulk mixing (tick marks are 5% increments) aim to model assimilation of low-␦18O and high-␦18O crusts, and addition of high-␦18O fluids to magma generation zones. All fluids and assimilants have 87Sr/86Sr ϭ 0.7045; numbers in ovals are ␦18O for fluids, numbers in squares are for assimilants. The data is most consistent with derivation of both high- and low-␦18O magmas from similar high-␦18O and low-␦18O crustal sources, rather than a source enrichment process; the latter would require an unrealistically high proportion (ϳ10%) of fluid with ␦18O(fluid) ϭϩ20‰, and low 87Sr/86Sr (ca. Ͻ 0.704). See text for discussion. (B) Basic volcanic rocks (SiO2 59%). With few exceptions (see Table 1) basic rocks tend to form a positive trend towards high-␦18O values; silicic rocks scatter and many have low-␦18O magmatic values. This is in contrast with other circum Pacific arcs: Tonga-Kermadec (T-K, Bindeman, Turner et al., in prep) and Mariana arcs (shaded area, data are from fig. 3 of Eiler et al., 2000) (C, D) Rocks from the Central Kamchatka Depression (CKD, this work and published sources). Normal-␦18O array for high-Mg basalt is from Figure 4. Notice initially high-␦18O character of basic rocks and significant initial scatter in both ␦18O and 87Sr/86Sr isotopic ratios. Arrows in D aim to explain isotopic variation by mixing of lower crustal (from Fig. 3B), arc-mantle source (AMS), and a slab-melt adakitic component (low-Sr/Y, low87Sr/86Sr, high-Mg-andesite, e.g., Drummond et al., 1996). Volcanic arc of Kamchatka 857

Fig. 9. Radiogenic isotopes of Nd and Pb vs. ␦18O(magma) (A–C). Data are from Tables 1 and 3. Note (A–C) rather primitive parameters for composition of other radiogenic isotope systems and subtle negative trends for Pb isotopes, as volcanoes in the CKD (Shiveluch, Bezymianny) tend to have lower values of 206Pb/204Pb, and 208Pb/204Pb. MORB and Kurile island arc volcanic rocks (data from Bindeman and Bailey, 1999) are shown for comparison; Kamchatka is characterized by more variable and higher 87Sr/86Sr, but similar Pb and Nd isotopic ratios. D-F: 143Nd/144Nd vs. 87Sr/86Sr, and 207Pb/204Pb vs. 206Pb/204Pb, 208Pb/204Pb vs. 206Pb/204Pb diagrams showing differences between CKD and EVF volcanics, and potential mantle and crustal reservoirs that are capable of causing the isotopic variations. Sediment data is from this study (Table 4); crustal rocks are from Vinogradov (1995) and L’vov et al. (1986). 858 I. N. Bindeman et al.

Table 4. Ocean floor materials in front of the Kamchatka trench.

Sample* Iithology ␦18O, ‰ 87Sr/86Sr 143Nd/144Nd 206Pb/204Pb 207Pb/204Pb 208Pb/204Pb

Dredged sediments V21-151 mud 10.41 0.70588 0.51276 18.668 15.616 38.627 RAMA-30-Bx mud 11.47 0.70512 0.51278 RAMA-31-PG mud 13.19 0.70601 RAMA-31-P mud 11.05 0.70566 0.51272 RAMA-44-P mud 11.55 0.70698 0.51251 RAMA-45-G mud 11.73 0.70701 0.51251 RAMA-32-G mud 11.76 0.70641 0.51253 DSDP 192 192-32-2 clay 19.56 0.70786 192-24-2 clay 17.90 0.70651 192A-4-2 clay 21.31 0.70443 192-20-1 ooze 15.22 0.70624 192-6-2 diatom ooze 16.74 0.70719 192A-5-3 altered basalt 9.73 0.70393 0.51302 18.958 15.556 38.158 192-15-3 diatom ooze 23.35 0.70689 192-6-2 diatom ooze 9.93 0.70402

* See Bailey (1993, 1996) for sample description, and additional analyses of DSDP holes 193, 303, 304.

extensional tectonic regime and thinner crust in the CKD. 6. DISCUSSION Existing isotopic analyses of lower crustal rocks (Figs. 9D–F 18 and 3B) are permissive end-members for derivation of basic to 6.1. High-␦ O Basic Rocks of Kamchatka intermediate rocks through the interaction of arc mantle magma 18 with the lower crust by a MASH-type process (e.g., Hildreth The widespread occurrence of high-␦ O basalts and basaltic and Moorbath, 1988); this hypothesis is pending better support andesites in Kamchatka could be explained by: (1) arc mantle- from Os isotopes in mafic volcanics. like (ϩ5.8‰) basaltic magmas subsequently undergoing mod- erate amounts of assimilation of high-␦18O (ca. ϩ10‰) supra- 5.4. Comparison with Other Island Arc Systems crustal rocks in the lower or middle crust; (2) addition of high-␦18O (ca. ϩ10–20‰) slab fluids into magma generation A number of regional laser fluorination studies of O-isotopes zones; and (3) parental basic magmas with high-␦18O of 6.5 to in phenocrysts have recently been performed on other arcs 7.2‰, derived by melting of the thickened lower crust. (Thirlwall et al., 1996; McPherson et al., 1998; Eiler et al., In general, it is likely that all these sources variably contrib- 2000b; Vroon et al., 2001), and the results were correlated to uted to the overall high-␦18O signature. Either source (slab) or radiogenic isotopes and trace elements. These studies consid- lower to midcrustal metamorphic wall rock assimilants are ered both intraoceanic ensimatic (e.g., Marianas), and ensialic predominantly high-␦18O rocks, and should lead to increasing (e.g., Banda) arcs, but demonstrated only subtle magmatic ␦18O ␦18O in magmas. The abundance of lower-crustal and mantle variations in magmas within normal-␦18O range. Most previous xenoliths in high-␦18O volcanoes of CKD, and in Bakening and whole-work studies erroneously overestimated the ␦18O enrich- Avachinsky (EVF) (e.g., Widom et al., 2003) argues for a lower ment as a result of the secondary low-T alteration. For example, crustal to mantle-derived high-␦18O signature. Widom et al. Vroon et al. (2001) found only normal- to slightly high-␦18O (2003) interpreted relatively radiogenic 187Os/188Os (0.123 to range of 5.5 to 6.5‰ in ensialic Banda arc, in sharp contrast to 0.157) in metasomatized peridotite xenoliths as evidence for previous whole-rock studies that showed 5 to10‰ range (NB: multi-stage fluid addition to the arc mantle. Magma contami- there are no low-␦18O magmas, i.e., Ͻ5.5‰, in the Banda arc). nation by materials from the lower crust is supported by abun- Unlike Kamchatka, narrow ␦18O range in Banda’s volcanics, as dant and variously digested lower crustal amphibole-bearing well as even narrower ␦18O ranges in the Mariana and Tonga- xenoliths observed by us in Shiveluch, Bezymianny, and Kermadec arcs (Fig. 8b), are coupled with significant range in Avachinsky volcanoes; their host lavas tend to have high-␦18O radiogenic isotopes (e.g., 87Sr/86Sr ϭ 0.7045 to 710); these and 87Sr/86Sr (Fig. 9), similar to Mesozoic to Cenozoic lower- features at Banda arc are interpreted to largely represent con- crustal xenoliths elsewhere (e.g., Ducea, 2002, and references tribution from subducted continental material. In most other therein). For siliceous rocks of Khangar volcano (and older arcs (e.g., Fig. 8b), the narrow and MORB-like phenocrysts- volcanoes on the on the Sredinny Range), assimilation of based ␦18O values were attributed to reflect 0.1–2% fluid Ϯ high-␦18O and 87Sr/86Sr crustal rocks is documented (Dril et sediments addition to the mantle wedge, negligibly affecting al., 1999; Fig. 3A). This particular scenario should be con- bulk ␦18O. Where more sediments is subducted within a frac- strained by detailed petrogenetic studies at each volcanic cen- ture zone, slightly elevated ␦18O values may result (e.g., 5.8– ter. There is no unambiguous regional correlation of 6.5‰ at Seguam island in the Aleutians, Singer et al., 1992). ␦18O(magma) for basalt with distance from the trench, or with The discussion below presents compelling evidence for chiefly distance or position along the arc. crustal origin of strong ␦18O variations in the Kamchatkan In discriminating between models 1–3 above, we have at- volcanic arc. tempted to explain the mass balance of oxygen and Sr isotopes Volcanic arc of Kamchatka 859

Fig. 10. Correlation of ␦18O(magma) and 87Sr/86Sr with (A, B) crustal thickness and (C, D) slab depth (crustal and subduction parameters are from Bogdanov and Khain, 2000) for Kamchatkan volcanic arc. Notice: (1) significant scatter at each slab depth and crustal thickness; (2) scattered positive correlations for mafic rocks from EVF to CKD and SR, and worse correlation for silicic rocks (A, C); (3) Greater slab depth correlates positively with increasing crustal thickness. The increasing 87Sr/86Sr with slab depth (D) is the opposite to of the Kurile island arc (sloped box, data from Bindeman and Bailey, 1999), a southern extension of the same subduction zone; there is no dependence of ␦18O(magma) on slab depth for the Kuriles. The S and O isotopes in Kamchatka are interpreted to represent greater source variability and crustal influence on the isotopic record (see text). by performing simple bulk mixing calculations involving fluids wt%) of fluids are involved. Additionally, such model fluid and solid end-members with an expected range of values. Fluid should have an extremely high-␦18O(Ͼϩ20‰) with very contained 600 ppm Sr, 89% oxygen; assimilant and initial melt moderate 87Sr/86Sr ϭ 0.704 (Fig. 8A). This is difficult to contained 200–500 ppm Sr and 50% oxygen; ␦18O(fluid or achieve from the altered oceanic crust (␦18O ϭϩ2toϩ10‰, assimilant) ϭϩ10‰,15‰,20‰; 87Sr/86Sr ϭ 0.703 (initial Muehlenbachs, 1986), and is contrary to both ␦18O and 87Sr/ basalt), 0.7045 to 0.706 (assimilant or fluid), and ␦18O(upper 86Sr values in sediments (Table 4) that are dominated by crustal assimilant) were ranged from Ϫ15‰ to Ϫ5% and 0‰ arc-derived clastic materials with an insignificant proportion of (Fig. 8A). A bulk mixing model was chosen for simplicity since intraoceanic material. A carbonate-derived fluid (with it provides minimum proportions for fluid or assimilant addi- ␦18OϾ25‰) would satisfy end-member restrictions, but no tion. Classical or energy-constrained AFC should yield (pro- significant carbonate layers have been identified by dredging or hibitively) larger proportions of assimilant or fluid per mass of drilling in the NW Pacific (Bailey, 1993, 1996; Prueher and residual melt remaining. Rea, 2001). The fluid derived from Kurile-Kamchatka trench Slab fluid addition into mantle magma sources (i.e., slab- sediments is expected to be higher in 87Sr/86Sr (Ͼ0.706) and induced melting) underneath each volcano (e.g., Dorendorf et lower in ␦18O(ϩ8–10‰) than the required hypothetical end- al., 2000, for Kluychevskoy) with variable ␦18O(fluid) and member fluid. 87Sr/86Sr(fluid) composition can match the isotopic composi- Assimilation of crustal materials by mantle-equilibrated ba- tion of O and Sr in lavas only if excessive amounts (10–20 saltic magmas would require high 20–40% assimilant addition 860 I. N. Bindeman et al.

(Fig. 8B). If equal amounts of crystallization per mass of with Ͼ100°C hotter subarc mantle than previously thought assimilant added are involved, these proportions are twice as (Kelemen et al., 2002). Geophysical data on crustal structure high. Following the argument of Bindeman and Valley (2003), under CKD volcanoes (Fedotov and Masurenkov, 1991) are it would be even more difficult to achieve tens of percent consistent with a significant (ca. 10 km) transition zone near the assimilation when energy conservation (Spera and Bohrson, Moho which can be interpreted as remobilized arc roots. 2001) is considered. Thus, mass balance prohibits assimilation and, as with fluids, this process would be difficult to reconcile 6.2. Large-Scale 18O Depletion of Kamchatkan Crust and with the observed isotopic trends. Low-␦18O Magmas The simplest explanation of high-␦18O in basic magmas (Fig. 8B), the rather primitive radiogenic isotopic signatures, and the The outstanding result of the present study is the discovery observed V-shaped pattern of ␦18O(magma) vs. 87Sr/86Sr (Fig. that Kamchatka has a great abundance of low-␦18O magmas, 8A) is that of direct derivation of Kamchatkan magmas from comparable only to Iceland. Low-␦18O magmas evidence parental sources that are high and variable in ␦18O. In this case of magma interaction (through assimilation or bulk melting) magmas are formed by melting of older high-␦18O, or low- with even stronger 18O-depleted upper crustal country rocks. ␦18O volcanic arc rocks with moderate 87Sr/86Sr and 206Pb/ Accordingly, ␦18O values in low-␦18O magmas serve as an 204Pb in a high heat flux environments. The recycling of older upper estimate of crustal ␦18O values. Assimilation is viewed (Cretaceous to Cenozoic) volcanic materials (e.g., high-␦18O as melting of country rocks followed by mixing with the main hydrated metabasalts) provides a viable solution. High-␦18O differentiating magma reservoir. values in these sources could either be explained by long-term, The inferred degree of assimilation or melting is dependent deep-seated ␦18O-enrichment by slab fluids during various sub- on the average ␦18O value of the wall rocks, and this is a largely duction configurations; or by low-temperature near-surface al- unknown parameter. For example, if the assimilation propor- teration, followed by burial by tectonic and magmatic pro- tion were independently constrained at 20%, then the wall-rock cesses. The range of Ͼ 1‰ observed in olivines from (Kamchatkan crust) has Ϫ5toϪ10‰ (e.g., Fig. 8D), but is Klyuchevskoy volcano by Dorendorf et al. (2000) is interpreted equal to the assimilant if magmas are derived by bulk melting. here to suggest derivation by mixing of (at least) two sources: Hydrothermally-altered rocks were analyzed from surface an arc mantle source basalt, and high-␦18O, high-87Sr/86Sr exposures and boreholes of modern geothermal systems (Taran lower crustal melts. Large ␦18O and 87Sr/86Sr heterogeneity, et al., 1988; Zolotarev et al., 1999) and yielded low ␦18O values and the adakitic trace elemental signature that is seen in other down to Ϫ1‰. The ␦18O values of hydrothermally-altered volcanoes of the CKD, and in Shiveluch in particular (Figs. rocks in a 1500 m deep drillhole through the modern Mut- 8C,D) would require an additional adakitic component (mod- novsky geothermal system (100–300°C) indicate a moderate erately high O and low Sr isotopes). No ␦18O characterization level of depletion down to 0–3‰ (Taran et al., 1988), in for adakite exists, but it is assumed here to be moderately equilibrium with moderately low values of Ϫ10 to Ϫ14‰ in elevated, since adakitic melts are thought to be derived from the present-day meteoric waters (Cheshko, 1994). However, most hydrated slab ( Ϯ sediments) which tends to be high in ␦18O. potential assimilants for genesis of Pleistocene and early Ho- The majority of modern volcanic rocks in Kamchatka (Figs. locene low-␦18O magmas have been affected by older fossil 18 8, 9) are intermediate in SiO2, whereas primitive basalts are hydrothermal systems. In the latter, the level of O-depletion rare in Kamchatka and in similar arcs built on thick crust (e.g., is stronger, since these rocks interacted with extremely light Andes; Hildreth and Moorbath, 1988). An origin of the domi- (ϽϪ25‰) Pleistocene glacial melt waters, and presumably the nant and voluminous magma types of basaltic andesites and geothermal systems died out or were quenched by glaciers andesites by crustal remelting can be plausibly driven by the during the Pleistocene. Rocks in equilibrium with these waters heat of primitive basalts and high heat flux. Therefore, our may be variably depleted down to Ϫ15‰, and ␦18O-depletion preferred model for the majority of Kamchatkan basic magmas of 2‰ in a magma would require only a moderate (Ͻ10–20%) is that of recycling and remobilization of high-␦18O Kamchat- amount of assimilation. kan arc roots. Such a model of multi-stage enrichments, stor- Subtle negative correlation of ␦18O with radiogenic isotopes age, tectonic and volcanic burial, and remelting is supported by of Sr (and Pb, Figs. 8A, 9B,C) supports shallow recycling of the low level of Th-disequilibria in Kamchatkan volcanic rocks older (Quaternary and older Cenozoic) and slightly more ra- (Turner et al., 1998), which show absent or small (ca. “Ͻ150 diogenic material. The latter is likely due to both build up of k.y.”) disequilibria, and do not exhibit significant Ra or 232Th radiogenic Sr due to aging effects in low-␦18O crust, as well as excesses. More significant U-Th disequilibria are seen in vol- leaching, transport, and addition of radiogenic Sr by circulating canoes of CKD (Dosseto and Bourdon, 2002) and suggests that hydro thermal solutions from older and more radiogenic crustal mantle melting, assimilation and mixing were sufficiently rapid rocks (87Sr/86Sr ϭ 0.706–0.710, Fig. 3B) since low-␦18O (e.g., Ͻ100 k.y.); we speculate that some of the U-Th disequi- geothermal waters tend to be high in 87Sr/86Sr (Vinogradov and libria may have been generated during devolatilization melting Vakin, 1983). of lower crustal amphibole-bearing metabasalts underplated by hot mantle-derived magmas (e.g., Rapp and Watson, 1995). 6.3. Climate Influence on Magma Sources? Slab detachment beneath CKD (e.g., Levin et al., 2003) may cause rather rapid upwelling of deep hot mantle and interaction Climate modeling studies suggest a wetter and colder climate with the mafic arc roots, yielding voluminous volcanism, some for the NW Pacific in the Pleistocene (Jouzel et al., 1994; with intraplate affinity. Additionally, recent treatments of ther- Kutzbach et al., 1998), in agreement with the palynological mal regime in the convecting mantle wedge are more consistent record on land (Braitseva et al., 1968) and in sediments (Ber- Volcanic arc of Kamchatka 861

Fig. 11. Primary ␦18O(magma) values (calculated from phenocrysts) plotted against age of eruption for Kamchatka. The lower ␦18O values for Pleistocene likely reflect assimilation of wall rocks that have interacted with low-␦18O glacial meltwaters. Holocene volcanics show a larger scatter and trend towards higher values, because Holocene magmas assimilated both Pleistocene (low-␦18O) and Holocene (higher-␦18O) crust. Age on Pleistocene volcanic units is either determined by fission tracks in apatite, short-lived magnetic reversals, or geological data (compiled after Sheymovich, 1979; Melekestsev et al., 1988; Grib and Leonov, 1993a,b; Leonov and Grib, 1998; V. L. Leonov, personal communication); uncertainty on Ͼ20 ka volcanics is large and their ages relative to glacial/interglacial are not yet certain. Ice Volume curve (solid line) is from Bradley (2000). ingia Drilling Project, 2001). Lower sea level stands created ␦18O crust. The persisting and abundant low-␦18O magmas of conditions for glaciations downwind (inland) on the eastern and the Holocene record a “memory” effect of Ͼ2 m.y.-long Pleis- south-eastern margins of north Pacific arcs (Mann and Peteet, tocene glaciation: many Holocene magmas could be derived 1994). Collectively, higher precipitation levels, colder climate, from synglacially altered crust, or reflect its assimilation. and rainout effects behind glaciers promoted voluminous pre- The great abundance (by number of volcanic units and their cipitation of low-␦18O and low-␦D meteoric water inland in total volume) of low-␦18O magmas in Kamchatka can be com- Kamchatka. The exact isotopic values during the Pleistocene pared to the Pleistocene-Holocene record in Iceland. Kam- were likely to vary temporally and laterally, but on average chatka was the second most volcanically active subaerial region they were substantially lower than today. The ␦18O values of during the Holocene (Simkin and Siebert, 1994); high endo- precipitation were likely below Ϫ25‰ during the last glacia- genic heat flux, and 18O-depleted waters are responsible for the tion, by comparison with other glaciated areas (e.g., Dansgaard large-scale depletion of the upper crust of Kamchatka (and et al., 1993). Although the continental ice sheet during the last Iceland) with respect to 18O. Widespread 18O-depleted areas glaciation in Kamchatka is being dabated (e.g., Grosswald, are described from Mesozoic terrains and attributed to their 1998), voluminous and interconnected alpine glaciers which subpolar location (e.g., Blattner et al., 1997). Identification of survived from winter to winter would be enough to yield the large Late Cenozoic province of Kamchatka is a significant ultra-low (ca. ϽϪ35‰) ␦18O snow meltwaters and cause a discovery which can serve as a model province for the inter- regional low-␦18O fingerprint in rocks and magmas. The abun- pretation of past ␦18O records preserved in rocks. We suggest dant low-␦18O magmas provide a record of this widespread 18O that earlier colder periods in the Earth’s history (e.g., Rumble depletion of the upper crust in Kamchatka, and can be attrib- and Yui, 1998) such as the “snowball Earth,” when moderately uted to climate change during Pleistocene-Holocene. The ultra- low-␦18O glaciers covered the surfaces of entire land-masses at low ␦18O meteoric waters that originated from these glaciers low latitudes, as well as the subpolar position of some conti- fed the geothermal systems and imparted a regional low-␦18O nents (with ultra-low-␦18O glaciers) at certain geologic periods, fingerprint to rocks and magmas and thereby affected ␦18O and can be inferred from the abundance of low-␦18O magmas both 87Sr/86Sr values of recent volcanic rocks. There has been a regionally and world-wide. single report of exceptionally low-␦18O(Ϫ28‰), and low-␦D Ϫ ( 130‰) waters encountered in the borehole near Mutnovsky 6.4. Implications for Other Convergent Margins volcano (Pokrovsky, 2001). Such low values are lower than ␦18O in the present Kamchatkan glaciers (ca. Ϫ20) and, if not We were able to achieve discrimination of crustal vs. mantle the result of complex underground evaporation-distillation pro- processes in Kamchatka, largely because of the oxygen isotopic cesses, they might represent glacial water that has survived diversity of Kamchatkan magmas, possibly fingerprinted by the from the last glacial maximum. effect of the last glaciation. In lower latitude arcs, oxygen When plotted against age (Fig. 11), primary magmatic ␦18O isotopic variations were shown to be more subdued (e.g., Baker values show a significant range in the Holocene, but are gen- et al., 1996; Eiler et al., 2000b; Vroon et al., 2001; our unpub- erally low in ␦18O for the Pleistocene. Low-␦18O values in lished data for Tonga-Kermadec), as the isotopic contrast be- Pleistocene magmas can be attributed to lower ␦18O values of tween ␦18O of Pleistocene precipitation and ␦18O(magma) was meteoric waters and associated hydrothermal systems. Normal less severe. The torque of oxygen isotope geochemistry is thus to high-␦18O values of Holocene magmas reflect new magmas less powerful. Only in Alaska-Aleutian arc is there a compa- that did not interact with either Pleistocene or Holocene low- rable, but smaller range in ␦18O values in magmas (4.5 to 862 I. N. Bindeman et al.

6.2‰, Singer et al., 1992; Bindeman et al., 2001); more syn- Bailey J. C. (1993) Geochemical history of sediments in the north- glacial volcanoes/calderas need to be sampled there. Nonethe- western Pacific Ocean. Geochem. J. 27, 71–91. less, similar processes of shallow and deep recycling are likely Bailey J. C., Larsen O., and Frolova T. I. (1987) Strontium isotope variations in Lower Tertiary-Quaternary volcanic rocks from the to be characteristic for other, lower latitude arcs, but the iso- Kurile island arc. Contrib. Mineral. Petrol. 95, 155–165. topic effect on magmas may not be currently analytically Baker J. A., Macpherson C. G., Menzies M. A., Thirlwall M. F., detectable. In this respect Kamchatka represents the best ex- Al-Kadasi M., and Mattey D. P. (2000) Resolving crustal and mantle ample (“natural laboratory”) to clearly demonstrate such pro- contributions to continental flood volcanism, Yemen: Constraints ␦18 from mineral isotope data. J. Petrol. 41, 1808–1820. cesses. The large-scale low- O anomaly in the upper crust of Baranov B. V., Seliversov N. I., Muraviev A. V., and Mazurov E. L. Kamchatka contradicts earlier reports that the measured D/H, (1991) The Kommandorsky Basin as a product of spreading behind 18O/16O, and87Sr/86Sr ratios in volcanics and their phenocrysts a transform plate boundary. Tectonophysics 99, 237–269. in many Kamchatkan volcanoes directly reflect those in the Beget J. E. (2001) Continuous Late Quaternary proxy climate records mantle (e.g., Volynets, 1994; Taran et al., 1997; Pineau et al., from loess in Beringia. Quat. Sci. Rev. 20, 499–507. Beringia Drilling Project. (2001) Holocene paleoclimate data from the 1999; Pokrovsky and Volynets, 1999; Churikova et al., 2001). Arctic: Testing models of global climate change. Quat. Sci. Rev. 20, The difference in ␦18O values in differentiated volcanics be- 1275–1287. tween neighboring calderas may reach 3‰ (e.g., Uzon vs. Bindeman I. N. and Bailey J. C. (1994) A model of reverse differen- Bolshoi Semiachik, Table 2) and, as we have shown above, is tiation at Dikii Greben’ Volcano, Kamchatka: Progressive basic ␦18 ␦ magma vesiculation in a silicic magma chamber. Contrib. Mineral. due to near-surface processes. The O and D values of Petrol. 117, 263–278. modern fumarolic and geothermal waters in Kamchatka are Bindeman I. N. and Bailey J. C. (1999) Trace elements in anorthite generally lower than in the neighboring Kurile islands (Giggen- megacrysts from the Kurile island arc: A window to across-arc bach, 1992; Cheshko, 1994; Taran et al., 1997; Fisher et al., geochemical variations in magma compositions. Earth Planet. Sci. 1998), which is consistent with latitudinal and altitudinal ef- Lett. 169, 209–226. Bindeman I. N., Fournelle J. H., and Valley J. W. (2001) 9100 BP fects on meteoric water precipitation, and on a shallow rather Eruption of Fisher Caldera, Unimak, Aleutians: Low-␦18O zoned 18 than deep origin of ␦ O and ␦D variations. tuff, a tephrochronological marker. J. Volc. Geotherm. Res. 111, We presented arguments above that both high- and low-␦18O 35–53. signatures are of crustal origin and best explained by lower- Bindeman I. N., Vinogradov V. I., Valley J. W., Wooden J. L., and Natalin B. A. (2002) Archean protolith, and accretion of crust in and upper- crustal processes, respectively. This study, based on 18 Kamchatka: SHRIMP dating of zircons from metamorphic rocks of ␦ O, also suggests the indirect influence of climate on isotopic Sredinny and Ganal Massifs. J. Geol. 110, 271–289. signatures of volcanism, and demonstrates the importance of Bindeman I. N. and Valley J. W. (2003) Rapid generation of both high- recycling older arc-derived crustal materials in recent postgla- and low-␦18O, large volume silicic magmas at the Timber Mountain/ cial volcanism; this has significant implications for lithosphere- Oasis Valley caldera complex, Nevada. Geol. Soc. Am. Bull. 115, 581–595. scale oxygen isotope stratification in modern and ancient con- Blattner P., Grindley G. W., and Adams C. J. (1997) Low-18O terranes vergent margins, in newly created (Iceland) and newly accreted tracking Mesozoic polar in the South Pacific. Geochim. (Kamchatka) lithospheres. Cosmochim. Acta 61, 569–576. Bogdanov N. A. and Khain V. E. (eds.) (2000) Tectonic map of the region. Explanatory notes. Scale 1:2,500,000. Nedra. Acknowledgments—The results of this work were made possible by the Bradley R. S. (2000) Paleoclimatology: Reconstructing Climates of the decade-long field efforts of the authors. We are grateful to O. A. Quaternary 2nd ed. Academic. Braitseva, B. G. Pokrovsky, O. B. Selyangin, M. M. Pevzner, T. I. Braitseva O. A., Melekestsev I. V., Evteeva I. C., and Lupikina E. G. Frolova, I. A. Burikova, V. L. Leonov, E. N. Grib, P. Izbekov, Philip (1968) Stratigraphy of Quaternary deposits and glaciations of Ka- Kyle and Peter Rinkleff for sample donation and discussion. Several mchatka. (in Russian).Nauka. important samples studied in this paper originally came from the Braitseva O. A., Litasova S. N., and Ponomarenko A. K. (1987) collections of the late O. N. Volynets, handled by A. B. Perepelov. Application of tephrochronological method for dating of the key Discussions with John Fournelle on the Aleutians helped us with this archaeological site in Eastern Kamchatka. Volcanol. Seismol. 5/5, Kamchatka study; early work by V. I. Vinogradov in Kamchatka 507–514. provoked many of the interpretations favored here. Mike Spicuzza is Braitseva O. A. and Melekestsev I. V. (1990) Eruptive history of thanked for assistance in stable isotope analyses. Comments and re- Karymsky volcano, Kamchatka, USSR, based on tephra stratigraphy views by Chris Harris, Greg Holk, Maxim Portnyagin, Bill Leeman, and 14C dating. Bull. Volcanol. 53, 195–206. Simon Sheppard, and an anonymous reviewer are greatly appreciated. Braitseva O. A., Melekestsev I. V., Ponomareva V. V., and Suler- We thank DOE (93ER14389) for funding the oxygen isotope research. zhitsky L. D. (1995) The ages of calderas, large explosive craters and Field work was funded by grants 6215–98 and 6543–99 from the active volcanoes in the Kuril-Kamchatka region, Russia. Bull. Vol- National Geographic Society and by the Russian Foundation for Basic canol. 57/6, 383–402. Research. Braitseva O. A., Melekestsev I. V., Ponomareva V. V., and Kirianov V. Y. (1996) The caldera-forming eruption of Ksudach volcano Associate editor: S. M. F. Sheppard about AD 240, the greatest explosive event of our era in Kamchatka. J. Volcanol. Geotherm. Res. 70, 49–66. REFERENCES Braitseva O. A., Ponomareva V. V., Sulerzhitsky L. D., Melekestsev I. V., and Bailey J. C. (1997a) Holocene key-marker tephra layers in Anderson A. T., Clayton R. N., and Mayeda T. (1971) Oxygen isotope Kamchatka, Russia. Quat. Res. 47, 125–139. thermometry of mafic igneous rocks. J. Geol. 79, 714–729. Braitseva O. A., Sulerzhitsky L. D., Ponomareva V. V., and Melekest- Appora I., Eiler J. M., Matthews A., and Stolper E. M. (2003) Exper- sev I. V. (1997b) Geochronology of the greatest Holocene explosive

imental determination of oxygen isotope fractionations between CO2 eruptions in Kamchatka and their imprint on the Greenland glacier vapor and soda-melilite melt. Geochim. Cosmochim. Acta 67, 459– shield. Trans. (Doklady) Russian Acad. Sci. Earth Sci. Sect. 352, 471. 138–140. Bailey J. C. (1996) Role of subducted sediments in the genesis of Brigham-Grette J. (2001) New perspectives on Beringian Quaternary Kurile-Kamchatka island arc basalts: Sr isotopic and elemental ev- paleogeography, stratigraphy, and glacial history. Quat. Sci. Rev. 20, idence. Geochem. J. 30, 289–321. 15–24. Volcanic arc of Kamchatka 863

Chacko T., Cole D. R., and Horita J. (2001) Equilibrium oxygen, dynamic model for the interpolation and extrapolation of liquid-solid hydrogen, and carbon isotope fractionation factors: Applicable to equilibria at elevated temperatures and pressure. Contrib. Mineral. geologic systems. Rev. Mineral. Geochem. 43, 1–82. Petrol. 119, 197–212 . Cheshko A. L. (1994) D, 18O/16O, and 3He/4He data on the formation Giggenbach W. F. (1992) Isotopic shifts in waters from geothermal and of the main types of thermal waters in the Kurile-Kamchatka region. volcanic systems along convergent plate boundaries and their origin. Geochem. Int. 32, 988–1001. Earth Planet. Sci. Lett. 113, 495–510. Chiba H., Chacko T., Clayton R. N., and Goldsmith J. R. (1989) Gill J. B. (1981) Orogenic Andesites and Plate Tectonics. Springer- Oxygen isotope fractionations involving diopside, forsterite, magne- Verlag. tite and calcite; application to geothermometry. Geochim. Cosmo- Gorbatov A., Fukao Y., Widiyantoro S., and Gordeev E. (2001) Seis- chim. Acta 53, 2985–2995. mic evidence for a mantle plume oceanward of the Kamchatka- Churikova T., Dorendorf F., and Woerner G. (2001) Sources and fluids Aleutian trench junction. Geophys. J. Int. 146, 282–288. in the mantle wedge below Kamchatka, evidence from across-arc Grib E. N. and Leonov V. L. (1993a) Ignimbrites of the Bolshoi geochemical variation. J. Petrol. 42, 1567–1593. Semiachik caldera, Kamchatka: Composition, structure and origin. Condomines M., Gronvold K., Hooker P. J., Muehlenbachs K., Volcanol. Seismol. 14, 532–550. O’Nions R. K., Oskarsson N., and Oxburgh E. R. (1983) Helium, Grib E. N. and Leonov V. L. (1993b) Ignimbrites of Uzon-Geyzernaya oxygen, strontium and neodymium isotopic relationships in Icelan- volcano-tectonic depression, Kamchatka: Correlation of sections, dic volcanics. Earth Planet. Sci. Lett. 66, 125–136. compositions and conditions of formation. Volcanol. Seismol. 14, Dansgaard W., Johnsen S. J., Clausen H. B., Dahljensen D., Gun- 15–33. destrup N. S., Hammer C. U., Hvidberg C. S., Steffensen J. P., Grosswald M. G. (1998) Late-Weichselian ice sheets in Arctic and Sveinbjornsdottir A. E., Jouzel J., and Bond G. (1993) Evidence for Pacific . Quat. Int. 45, 3–18. general instability of past climate from a 250-kyr ice-core record. Nature 364, 218–220. Gualtieri L., Glushkova O., and Brigham-Grette J. (2001) Evidence for Davidson J. P. (1989) Crustal contamination vs. subduction zone en- restricted ice extent during the last glacial maximum in the Koryak richment: Examples from the Lesser Antilles and implications for the Mountains of Chukotka, Far Eastern Russia. Geol. Soc. Am. Bull. mantle source composition of island arc volcanic rocks. Geochim. 112, 1106–1118. Cosmochim. Acta 51, 2185–2198. Harris C., Smith H. S., and le Roex A. P. (2000) Oxygen isotope Donnely-Nolan J. M. (1998) Abrupt shift in delta 18O values at Med- composition of phenocrysts from Tristan da Cunha and Gough Island icine Lake volcano (California, USA). Bull. Volcanol. 59, 529–536. lavas: Variation with fractional crystallization and evidence for Dorendorf F., Wiechert U., and Worner G. (2000) Hydrated sub-arc assimilation. Contrib. Mineral. Petrol. 138, 164–175. mantle: A source for the Kluchevskoy volcano, Kamchatka/Russia. Hattori K. and Muehlenbachs K. (1982) Oxygen isotope ratios of the Earth Planet. Sci. Lett. 175, 69–86. Icelandic crust. J. Geophys. Res. 87, 6559–6565. Dosseto A. and Bourdon B. (2002) Models for genesis of Kamchatka Hemond C., Condomines M., Fourcade S., Alle`gre C. J., Oskarsson N., arc magmas: New insights from U-series (abstract). Geochim. Cos- and Javoy M. (1988) Thorium, strontium and oxygen isotopic geo- mochim. Acta 66(15A), A193–A193. chemistry in recent tholeiites from Iceland: Crustal influence on Dril S. I., Pokrovsky B. G., Perepelov A. B., Baluyev E. Y. and mantle-derived magmas. Earth Planet. Sci. Lett. 87, 273–285. Bankovskaya E. V. (1999) Khangar Volcano (Kamchatka): 87Sr/ Hildreth W. and Moorbath S. (1988) Crustal contribution to arc mag- 86Sr-␦18O isotope systematics and genesis of lavas. In Proceedings matism in the Andes of Central Chile. Contrib. Mineral. Petrol. 98, of the International Conference on Isotope Geochemistry, p. 124. 455–489. Moscow. Vernadsky Institute (Abstr.). Hochstaedter A. G., Kepezhinskas P., and Defant M. (1996) Insights Drummond M. S., Defant M. J., and Kepezhinskas P. K. (1996) into the volcanic arc mantle wedge from magnesian lavas from the Petrogenesis of slab-derived trondhjemite-tonalite-dacite/adakite Kamchatka arc. J. Geophys. Res. 101, 697–712. magmas. Trans. R. Soc. Edinburgh 87, 205–215. Hoefs J. (1997) Stable Isotope Geochemistry 4th ed. Springer-Verlag. Ducea M. N. (2002) Constraints on the bulk composition and root Ishikawa T., Tera F., and Nakazawa T. (2001) Boron isotope and trace foundering rates of continental arcs: A California arc perspective. J. element systemetics of the three volcanic zones in the Kamchatka Geophys. Res. 107(B11), 2304. arc. Geochim. Cosmochim. Acta 65, 4523–4537. Eiler J. M. (2001) Oxygen isotope variations in basaltic lavas and upper Ivanov B. V. (1990) Types of Andesite Volcanism in the Pacific mantle rocks. Rev. Mineral. Geochem. 43, 319–364. Volcanic Belt (in Russian). Nauka. Eiler J. M., Gronvold K., and Kitchen N. (2000a) Oxygen isotope Izbekov P. E., Eichelberger J. C., Patino L. C., Vogel T. A., and Ivanov evidence for the origin of chemical variations in lavas from the B. V. (2002) Calcic cores of plagioclase phenocrysts in andesite Theistareykir volcano in Iceland’s northern volcanic zone. Earth from Karymsky volcano: Evidence for rapid introduction by basaltic Planet. Sci. Lett. 184, 269–286. replenishment. Geology 30, 799–802. Eiler J. M., Crawford A., Elliott T., Farley K. A., Valley J. W., and Jouzel J., Koster R. D., Suozzo R. J., and Russell G. L. (1994) Stable Stolper E. M. (2000b) Oxygen isotope geochemistry of oceanic-arc water isotope behavior during the last glacial maximum—A general- lavas. J. Petrol. 41, 229–256. circulation model analysis. J. Geophys. Res. Atm. 99, 25791– Engebretson D. C., Cox A. and Gordon R. G. (1985) Relative motions 25801. between oceanic and continental plates in the Pacific basin. Special Kalamarides R. I. (1984) Kiglapait geochemistry VI: Oxygen isotopes. Paper 206. Geol. Soc. Am. Erlich E. (1986) Geology of the calderas of Kamchatka and Kurile Geochim. Cosmochim. Acta 48, 1827–1836. islands with comparison to calderas of and the Aleutians, Kelemen P., Parmentier M., Rilling J., Meh L., and Hacker B. (2002) Alaska. Open File Report 86-291. U.S. Geological Survey. Thermal convection of the mantle wedge. Geochim. Cosmochim. Fedotov S. A. and Masurenkov Yu. P. (eds.) Active Volcanoes of Acta 66(15A), A389–A389. Kamchatka. (1991) 2 vols. Nauka. Kepezhinskas P., Defant M. J., and Drummond M. S. (1996) Progres- Fisher T. P., Giggenbach W. F., Sano Y., and Williams S. N. (1998) sive enrichment of island arc mantle by melt-peridotite interaction Fluxes of volatiles discharged from Kudriavy, a subduction zone inferred from Kamchatka xenoliths. Geochim. Cosmochim. Acta 60, volcano, Kurile Islands. Earth Planet. Sci. Lett. 160, 81–96. 1217–1229. Gautason B. and Muehlenbachs K. (1998) Oxygen isotopic fluxes Kepezhinskas P., McDermott F., Defant M. J., et al.. (1997) Trace associated with high-temperature processes in the rift zones of Ice- element and Sr-Nd-Pb isotopic constraints on a three-component land. Chem. Geol. 145, 275–286. model of Kamchatka arc petrogenesis. Geochim. Cosmochim. Acta Geist E. L., Vallier T. L., and Scholl D. W. (1994) Origin, transport and 61, 577–600. emplacement of exotic island arc terranes exposed in eastern Kam- Kersting A. B. and Arculus R. J. (1994) Klyuchevskoi volcano, Kam- chatka, Russia. Geol. Soc. Am. Bull. 106, 1182–1194. chatka, Russia: The role of high flux, recharged, tapped and frac- Ghiorso M. S. and Sack R. O. (1995) Chemical mass-transfer in tionated magma chamber(s) in the genesis of high Al2O3 from high magmatic processes IV: A revised and internally-consistent thermo- MgO basalt. J. Petrol. 35, 1–42. 864 I. N. Bindeman et al.

Kersting A. B. and Arculus R. J. (1995) Pb isotopic composition of study at 550–950°C and 1 bar. Geochim. Cosmochim. Acta 60, Klyuchevskoy volcano, Kamchatka and north Pacific sediments: 1963–1973. Implications for magma genesis and crustal recycling in the Kam- Pineau F., Semet M. P., Grassioneau N., Okrugin V. M., and Javoy M. chatka arc. Earth Planet. Sci. Lett. 136, 133–148. (1999) The genesis of the stable isotope (O, H) record in arc Kohn M. J. and Valley J. W. (1998) Oxygen isotope geochemistry of magmas: The Kamchatka’s case. Chem. Geol. 135, 93–124. the amphiboles: Isotope effects of cation substitutions in minerals. Plank T. and Langmuir C. H. (1998) The geochemical composition of Geochim. Cosmochim. Acta 62, 1947–1958. subducting sediment and its consequences for the crust and mantle. Konstantinovskaya E. A. (2000) Geodynamics of an Early Eocene Chem. Geol. 145, 325–394. arc-continent collision reconstructed from Kamchatka Orogenic Pokrovsky B. G. (2001) Crustal Contamination of Mantle Magmas by Belt, NE Russia. Tectonophysics 325, 87–105. Methods of Isotope Geochemistry (in Russian). Nauka. Kutzbach J., Gallimore R., Harrison S., Behling P., Selin R., and Laarif Pokrovsky B. G. and Volynets O. N. (1999) Oxygen-isotope geochem- F. (1998) Climate and biome simulations for the past 21,000 years. istry in volcanic rocks of the Kurile-Kamchatka arc. Petrology 7, Quat. Sci. Rev. 17, 473–506. 227–251. Leeman W. P. (1996) Boron and other fluid-mobile elements in vol- Ponomareva VV., Melekestsev I. V., Kyle P. R., Rinkleff P. G., canic arc lavas: Implications for subduction processes. In Subduc- Dirksen O. V., Sulerzhitsky L. D., Zaretskaia N. E. and Rourke R. tion: Top to Bottom. Geophysics Monograph 96, pp. 269–276. (2004) The 7600 (14C) year BP Kurile Lake caldera eruption, Ka- American Geophysics Union. mchatka, Russia: Stratigraphy and field relationships. J. Volcanol. Leonov V. L. and Grib E. N. (1998) Calderas and ignimbrites of Geotherm. Res., in press. Uzon-Semiachik area, Kamchatka. Volcanol. Seismol. 3, 41–59. Presnall D. C., Dixon T. H., O’Donell T. H., and Dixon S. A. (1979) Levin V., Shapiro N., Park J., and Ritzwoller M. (2003) Seismic Generation of mid-ocean ridge tholeiites. J. Petrol. 20, 3–35. evidence for catastrophic slab loss beneath Kamchatka. Nature 418, Prueher L. M. and Rea D. K. (2001a) of the Kam- 763–767. chatka-Kurile and Aleutian arcs: Evidence for volcanic episodicity. Lvov A. B., Neelov A. N., Bogomolov E. S., and Mikhailova N. S. J. Volcanol. Geotherm. Res. 106, 67–84. (1985) On the age of metamorphic rocks from Ganalsky Range, Prueher L. M. and Rea D. K. (2001b) Volcanic triggering of late Kamchatka. Geologia I Geophysika 7, 47–56. glaciation: Evidence from the flux of volcanic glass and Macias J. L. and Sheridan M. F. (1995) Products of 1907 eruption of ice-rafted debris to the North Pacific Ocean. Palaegeogr. Palaeocli- Shtubel volcano, Ksudach caldera, Kamchatka, Russia. Geol. Soc. matol. Palaeoecol. 173, 215–230. Am. Bull. 107, 969–986. Rapp R. P. and Watson E. B. (1995) Dehydration melting of metaba- Macpherson C. G., Gamble J. A., and Mattey D. P. (1998) Oxygen salts at 8–32 kbars: Implications for continental growth and crust- isotope geochemistry of lavas from an oceanic to continental arc mantle recycling. J. Petrol. 36, 891–931. transition, Kermadec-Hikurangi margin, SW Pacific. Earth Planet. Rinkleff P. G. (1999) Petrologic evolution and stratigraphy of the Sci. Lett. 160, 609–621. eruptive products from the 7.7ka (14C) Kurile Lake caldera eruption, Mann D. H. and Peteet D. M. (1994) Extent and timing of the last southern Kamchatka, Russia. M.S. thesis. New Mexico Institute of glacial maximum in SW Alaska. Quat. Res. 42, 136–148. Mining and Technology. Matsuhisa Y. (1979) Oxygen isotopic compositions of volcanic rocks Rumble D. and Yui T. F. (1998) The Qinglongshan oxygen and from the east Japan island arc and their bearing on petrogenesis. J. hydrogen isotope anomaly near Donghai in Jiangsu Province, China. Volcanol. Geotherm. Res. 5, 271–296. Geochim. Cosmochim. Acta 62, 3307–3321. Matsuhisa Y., Matsubaya O., and Sakai H. (1973) Oxygen isotope Ryan J. G., Morris J., Tera F., Leeman W. P., and Tsvetkov A. A. variations in magmatic differentiation processes of the volcanic (1995) Cross-arc geochemical variations in the Kurile arc as a rocks in Japan. Contrib. Min. Petrol. 39, 277–288. function of slab depth. Science 270, 625–627. Matthews A., Goldsmith J. R., and Clayton R. N. (1983) Oxygen Savoskul O. S. (1999) Holocene glacier advances in the headwaters of isotope fractionations involving pyroxenes: The calibration of min- Sredniaya Avacha, Kamchatka, Russia. Quat. Res. 52, 14–26. eral pair geothermometers. Geochim. Cosmochim. Acta 47, 631– Sheppard S. M. F. and Harris C. (1985) Hydrogen and oxygen isotope 644. geochemistry of Ascension Island lavas and granites: Variation with Matthews A., Palin J. M., Epstein S., Stolper and E. M. (1994) crystal crystallization and interaction with sea water. Contrib. Min- Experimental study of 18O/16O partitioning between crystalline al- eral. Petrol. 91, 74–81.

bite, albitic glass and CO2 gas. Geochim. Cosmochim. Acta 58, Sheymovich V. S. (1979) Ignimbrites of Kamchatka. Nedra. 5255–5266. Simkin T. and Siebert L. (1994) Volcanoes of the World. 2nd ed. Matthews A., Stolper E. M., Eiler J. M. and Epstein S. (1998) Oxygen Geoscience Press. isotope fractionation among melts, minerals and rocks. In Proceed- Singer B. S., O’Neil J. R., and Brophy J. G. (1992) Oxygen isotope ings of the 1998 Goldsmith Conference, Toulouse, pp. 971–972. constraints on the petrogenesis of Aleutian arc magmas. Geology 20, London Minerological Society. 367–370. Melekestsev I. V., Braitseva O. A., Erlich E. N., and Kozhemyaka Spera F. J. and Bohrson W. (2001) Energy-constrained open-system N. N. (1974) Volcanic mountains and plains (in Russian). In Kam- magmatic processes I: General model and energy constrained assim- chatka, Kurile and Commander Islands (ed. I. V. Luchitsky), pp. ilation-fractional crystallization (EC-AFC) formulation. J. Petrol. 162–234. Nauka. 42, 999–1118. Melekestsev I. V., Braitseva O. A., Sulerzhitsky and L. D. (1988) Stolper E. and Epstein S. (1991) An experimental study of oxygen

Catastrophic explosive volcanic eruptions in Kamchatka and the isotope partitioning between silica glass and CO2 vapor. Special Kurile islands in late Pleistocene-Early Holocene time. Trans. Publication 3, pp. 35–51. Geochemical Society. (Doklady) Acad. Sci. USSR 301, 55–59. Taran Yu. A., Pokrovsky B. G., and Volynets O. N. (1997) Hydrogen Muehlenbachs K. and Byerly G. (1982) 18O enrichment of silicic isotopes in amphiboles and micas from Quaternary volcanic rocks of magmas caused by crystal fractionation at the Galapagos spreading Kurile-Kamchatka arc system. Geochem. J. 34, 203–221. center. Contrib. Mineral. Petrol. 79, 76–79. Taran Yu. A., Pokrovsky B. G., and Glavatskikh S. F. (1988) Isotope Muehlenbachs K., Anderson A. T., and Sigvaldsson G. E. (1974) data on hydrothermal transformation of rocks in the Mutnovsky Low-␦18O basalts from Iceland. Geochim. Cosmochim. Acta 38, geothermal system, Kamchatka. Geochem. Int. 25, 1569–1579. 577–588. Tatsumi Y., Kogiso T., and Nohda S. (1995) Formation of a third Nakada M. and Yokose H. (1992) Ice age as a trigger of active volcanic chain in Kamchatka: Generation of unusual subduction- Quaternary volcanism and tectonism. Tectonophysics 212, 321–329. related magmas. Contrib. Mineral. Petrol. 120, 117–128. Nokleberg W. J., Parfenov L. M., Monger J. W. H., et al. (1998) Taylor H. P. Jr. (1968) Oxygen isotope geochemistry of igneous rocks. Phanerozoic tectonic evolution of the circum-north Pacific. Open Contrib. Mineral. Petrol. 19, 1–71. File Report 98-754. U.S. Geological Survey. Taylor H. P. Jr. and Sheppard S. M. F. (1986) Igneous rocks: I. Palin J. M., Epstein S., and Stolper E. M. (1996) Oxygen isotope Processes of isotopic fractionation and isotopic systematics. Rev.

partitioning between rhyolitic glass/melt and CO2: An experimental Mineral. 16, 227–272. Volcanic arc of Kamchatka 865

Tera F., Morris J. D., Leeman W. P., and Tsvetkov A. A. (1990) Volynets O. N., Babanskii A. D., and Gol’tsman Y. V. (2000) Varia- Further evidence from Be-B systematics for the homogeneity of the tions in isotopic and trace elemental composition of lavas from subducted component in arc magmatism: Case of the Kurile-Kam- volcanoes of the Northern group, Kamchatka, in relation to specific chatka arc. Geochronol. Isotope Geol. 27, 100. features of subduction. Geochem. Int. 38, 974–989. Thirlwall M. F., Graham A. M., Arculus R. J., Harmon R. S., and Vroon P. Z., Lowry D., Van Bergen M. J., Boyce A. J., and Mattey Macpherson C. G. (1996) Resolution of the effects of crustal assim- D. P. (2001) Oxygen isotope systematics of the Banda ilation, sediment subduction, and fluid transport in island arc mag- Arc: Low-␦18O despite involvement of subducted continental mas: Pb-Sr-Nd-O isotope geochemistry of Grenada, Lesser Antilles. material in magma genesis. Geochim. Cosmochim. Acta 65, 589– Geochim. Cosmochim. Acta 60, 4785–4810. 609. Turner S., McDermott F., Hawkesworth C., and Kepezhinskas P. Widom E., Kepezhinskas P., and Defant M. (2003) The nature of (1998) A U-series study of lavas from Kamchatka and the Aleutians: metasomatism in the sub-arc mantle wedge: Evidence from Re-Os Constraints on source composition and melting processes. Contrib. isotopes in Kamchatka peridotite xenoliths. Chem. Geol. 196, 283– Mineral. Petrol. 133, 217–234. 306. Valley J. W., Kitchen N., Kohn M. J., Niendorf C. R., and Spicuzza Yogodzinski G. M., Lees J. M., Churikova T. G., Dorendorf F., M. J. (1995) UWG-2, a garnet standard for oxygen isotope ratio: Woerner G., and Volynets O. N. (2001) Geochemical evidence for Strategies for high precision and accuracy with laser heating. the melting of subducting oceanic lithosphere at plate edges. Nature Geochim. Cosmochim. Acta 59, 5223–5231. Vinogradov V. I. (1995) Isotopic evidence of the conversion of oceanic 409, 500–504. crust to continental crust in the continent-ocean transition zone of Zhao Z.-F. and Zheng Y.-F. (2003) Calculation of oxygen isotope Kamchatka. Geochem. Int. 32, 70–109. fractionation in magmatic rocks. Chem. Geol. 193, 59–80. Vinogradov V. I. and Vakin Y. A. (1983) Strontium isotopes of the Zhuravlev D. Z., Tsvetkov A. A., Zhuravlev A. Z., Gladkov N. G., 143 144 87 86 thermal waters of Kamchatka. Trans. (Doklady) Acad. Sci. USSR and Chernyshev I. V. (1987) Nd/ Nd and Sr/ Sr ratios in 273, 965–968. recent magmatic rocks of the Kurile island arc. Chem. Geol. 66, Volynets O. N. (1994) Geochemical types, petrology and genesis of 227–243. Late Cenozoic volcanic rocks from the Kurile-Kamchatka island-arc Zielinski G. A., Mayewski Meeker L. D., Whitlow S., and Twickler M. system. Int. Geol. Rev 36/4, 373–405. (1996) A 110 000-year record of explosive volcanism from the Volynets O. N., Ponomareva V. V., Braitseva O. A., Melekestsev I. V., GISP2 (Greenland) ice core. Quat. Res. 45, 109–118. and Chen C. H. (1999) Holocene eruptive history of Ksudach vol- Zolotarev B. P., Karpov G. A., Yeroshev-Shak V. A., Artamonov canic massif, South Kamchatka: Evolution of a large magmatic A. V., Grigoryev V. S., and Pokrovsky B. G. (1999) Evolution of chamber. J. Volcanol. Geotherm. Res. 91, 23–42. volcanism in Uzon caldera. Volcanol. Seismol. 6, 67–84.